Visualizing and Interacting with Geospatial Networks: A Survey and Design Space

Abstract

This paper surveys visualization and interaction techniques for geospatial networks from a total of 95 papers. Geospatial networks are graphs where nodes and links can be associated with geographic locations. Examples can include social networks, trade and migration, as well as traffic and transport networks. Visualizing geospatial networks poses numerous challenges around the integration of both network and geographical information as well as additional information such as node and link attributes, time, and uncertainty. Our overview analyzes existing techniques along four dimensions: i) the representation of geographical information, ii) the representation of network information, iii) the visual integration of both, and iv) the use of interaction. These four dimensions allow us to discuss techniques with respect to the trade-offs they make between showing information across all these dimensions and how they solve the problem of showing as much information as necessary while maintaining readability of the visualization. https://geonetworks.github.io.

Full text

Volume xx (200y), Number z, pp. 1–27
Visualizing and Interacting with Geospatial Networks:
A Survey and Design Space
S. Schöttler1, Y. Yang2, H. Pfister2, and B. Bach1
1University of Edinburgh, UK
2Harvard University, MA
Abstract
This paper surveys visualization and interaction techniques for geospatial networks from a total of 95 papers. Geospatial
networks are graphs where nodes and links can be associated with geographic locations. Examples can include social networks,
trade and migration, as well as traffic and transport networks. Visualizing geospatial networks poses numerous challenges
around the integration of both network and geographical information as well as additional information such as node and link
attributes, time, and uncertainty. Our overview analyzes existing techniques along four dimensions: i) the representation of
geographical information, ii) the representation of network information, iii) the visual integration of both, and iv) the use of
interaction. These four dimensions allow us to discuss techniques with respect to the trade-offs they make between showing
information across all these dimensions and how they solve the problem of showing as much information as necessary while
maintaining readability of the visualization. https://geonetworks.github.io.
CCS Concepts
•Human-centered computing →Information visualization; Geographic visualization; Visualization techniques; Scientific
visualization; Graph drawings; •Computing methodologies →Shape modeling; Perception;
1. Introduction
Geospatial networks are graphs whose nodes and links can be asso-
ciated with geographic locations. Examples of geospatial networks
include social networks where social actors are found at specific lo-
cations, trade between countries (Fig. 1), or transport links between
defined locations (Fig. 2). In all these networks, nodes are associ-
ated with individual geographic locations such as a city, a country,
or a set of geographic coordinates, and are connected by links.
The visualization of geospatial networks goes back to at least
French civil engineer Charles Joseph Minard (1781–1870), fa-
mously known for his depiction of Napoleon’s March to Moscow
and numerous other flow map visualizations [Ren18]. Fig. 1shows
one of Minard’s graphics, visualizing the origin and amount of cot-
ton imported into Europe in 1858, 1864, and 1865. The width of the
flows shows the quantity of imported cotton and their color shows
the country of origin. Minard cleverly distorts geographic shapes,
positions, and sizes of countries, islands, and continents to provide
space for these links. Showing all three years juxtaposed allows for
understanding and exploring changes over time.
Another notable example of geospatial network visualization is
Harry Beck’s schematic map of the London Underground. De-
signed in 1933, Beck created his map (Fig. 2(b)) to solve the
problem of increasing complexity of the network. The growth in
lines and stations had made the traditional approach to transport
Figure 1: Charles Joseph Minard (1781-1870) depicting cotton im-
ported into Europe in 1858 (left), 1864 (center), and 1865 (right).
Color indicates the origin of flows: blue = United States; yellow =
India, Orient, China, etc.; brown = Egypt, Syria; violet = Brazil,
Oriental India; red = England, re-exportation.
maps (Fig. 2(a)), which was based on precise geographic locations,
harder to read and therefore unfit for public display. Beck noticed
that a lot of the geographic information in this map was unneces-
sary in the context it was to be used in—for tasks such as finding
the fastest route between two stations. This insight led Beck to dis-
tort the underlying map to display the network of transport lines
and stations more clearly; inspired by electronic circuit boards, he
submitted to COMPUTER GRAPHICS Forum (3/2021).
arXiv:2101.06322v2 [cs.HC] 24 Mar 2021

2S. Schöttler, Y. Yang, H. Pfister & B. Bach / Geospatial Network Visualization
(a) 1920 (b) 1933
Figure 2: Metro map designs in London: 1920 (left) showing accu-
rate spatial positions of stations; 1933 (right) showing a distorted
version by Harry Beck (1902-1974).
straightened lines and only used angles of 45 and 90 degrees. With
some exceptions and numerous extensions, Beck’s design has be-
come the standard solution for public transport maps, providing an
effective trade-off by abstracting geography to emphasize network
topology.
Today, numerous techniques exist to offer solutions to the in-
herent complexity in geospatial networks, posed by the combina-
tion of geographic with network and potentially other informa-
tion. Most of this work has been focusing on ‘traditional’ node-
link diagrams superimposed onto geographic maps. For example,
sometimes, links are drawn in a straight manner between source
and destination [e.g., BEW95], sometimes these links are slightly
curved [e.g., JSM*17]—perhaps with the intention to communicate
that the link does not have a specific geographic location. More re-
cent work has suggested automated routing approaches [BSV11a].
Virtual reality is also offering new ways for rendering and interact-
ing with three-dimensional globes [YDJ*19].
Another set of techniques continues Minard’s and Beck’s work
and automatically deforms geographic space [BDD*16;OHN*18].
In its most extreme form, geographic space gets abstracted almost
entirely and is instead represented as spatially grouped and ordered,
colored segments on a circle in a chord diagram [Hen13;AS14]
(Figure 6(b)). Alternatively to node-link diagrams, adjacency ma-
trices can solve the problem of dense networks side-by-side with
geographic maps [YDGM17;Guo07] and can group and order
nodes by geographic location if necessary [BPLL11;YDGM17].
Finally, there is a range of purely interactive techniques, e.g., to al-
low for navigating between distant nodes [MCH*09] or interactive
lenses to reduce local link clutter in node-link diagrams [WCG03].
In summary, the range of techniques is rich, and contributions
have come from many different communities: visualization, graph
drawing, geography. Moreover, many challenges are still unsolved.
Examples include moving nodes and dynamic geospatial networks
and uncertainty in network topology (e.g., missing nodes and
links), geographic locations (e.g., different granularities, identical
positions), and their combination (e.g., uncertain, multiple, or miss-
ing node and link positions).
While many surveys, books, and articles have been written
about visualizing networks, geographic visualization, and spatio-
temporal data visualization (see Section 2for an overview), there
is so far no structural approach to categorizing types of geospatial
networks and the respective techniques. Different and inconsistent
terminology, such as flow maps,origin-destination maps,geospa-
tial networks, etc., makes it hard to navigate the jungle of tech-
niques and to inform i) the application of existing techniques to
specific (domain) problems, ii) the design of novel techniques to
address open challenges, and iii) the comparison and study of the
effectiveness of a given set of techniques for a set of analysis tasks.
The goal of this survey is to provide a structured review and to
propose a design space for visualizations of geospatial networks,
as well as to inform a discussion about current challenges. While
our discussion of challenges focuses on how to practically address
specific attributes of geospatial networks, our design space is in-
formed by the trade-off each visualization design is confronted
with: emphasizing some information while abstracting and ag-
gregating other information in order to obtain a task-specific and
clearly readable visualization. This trade-off, nicely illustrated in
the works of Minard and Beck, is, in fact, common in visualiza-
tion design and is demonstrated by numerous studies that show the
complementariness of visualization designs.
In our design space (Section 3), we describe existing techniques
along four dimensions: i) GEO: how explicitly geographic infor-
mation is shown, ii) NET: how explicitly network information is
shown, iii) COMP: how geographic and network information are
composed and integrated visually, and iv) INTERACT: if and how
interactivity is used to facilitate integration and exploration.
This design space aims to capture the dimensions along which
a designer or analyst can make choices to create a balanced, pur-
poseful visualization design and to address specific visualization
challenges. This survey is addressed to readers in any discipline,
including students new to visualization or geography and their ap-
plications as well as to experts in any of these domains as a refer-
ence and design space.
This survey is structured as follows: After discussing related
work in Section 2, we provide definitions and terminology for
geospatial networks and their different types in Section 3. Section
4then details our methodology for finding, selecting, and coding
papers and how our design space evolved until its final state, while
Sections 5to 9explain our design space’s five dimensions and ex-
amples of visualization techniques. In Section 10, we discuss spe-
cific challenges and how they may be addressed. We conclude the
survey with a discussion and list of open problems in Section 11.
2. Related Work
Surveys on network visualization have been compiled for
many aspects in networks such as techniques for large graphs
[vLKS*11], group structures in graphs [VBW15], dynamic
graphs [BBDW14], multivariate networks [NMSL19], temporal
multivariate networks [AAK*14], multilayer networks [MGM*19],
and graph visualization in general [HMM00]. A variety of these
surveys include visualizations for geospatial networks but do so
for reasons other than surveying visualizations of geospatial net-
works as a whole. As a consequence, features and challenges spe-
cific to representing geospatial data are not discussed in detail, and
do not play a significant role in any taxonomies or classifications
introduced in these surveys. For example, surveys on edge bundling
techniques [ZPYQ13;LHT17a] frequently include techniques with
demonstrated applications to geospatial networks, or even specifi-
cally designed for this purpose, but they lack a wider discussion on
visualization of geospatial networks in general.
submitted to COMPUTER GRAPHICS Forum (3/2021).

S. Schöttler, Y. Yang, H. Pfister & B. Bach / Geospatial Network Visualization 3
Furthermore, a variety of surveys and textbooks on geographic
and spatio-temporal visualization have been published. Bertin
discusses maps as well as networks in his ‘Semiology of Graph-
ics’ [Ber83], but not the combination of the two. Cartography,
thematic mapping, and map design are discussed in numerous
works by cartographers [e.g., Rob95;DTH09;SMKH09;Fie18],
but networks play only a small role, if any at all, in these
books. A survey on ‘map-like’ visualization was presented by
Hogräfer et al. [HHS20], describing techniques that either imi-
tate or schematize cartographic maps in terms of their primary de-
sign elements: points, lines, areas, and fields. For spatio-temporal
data, Andrienko and Andrienko offer a systematic approach for ex-
ploratory analysis [AA06], Andrienko et al. discuss visual analytics
of movement data [AAB*13], and Bach et al. propose a descriptive
framework for spatio-temporal visualizations based on generalized
space-time cubes [BDA*17].
Geospatial networks have been discussed in a set of smaller
surveys, focusing mostly on node-link diagrams, graph drawing,
flow maps, trajectories [HCC*19] or specific applications such as
crime [Whe15] or climate [NBD*15]. Surveys on automatically
drawing schematic transit maps, a topic that has received con-
siderable attention in the graph drawing community, were pub-
lished in 2007 and 2020 [Wol07;WNT*20]. Rodgers [Rod05] pro-
vided a smaller overview of only node-link representations and
graph drawing techniques. Similarly, Wolff discussed the use of
graph drawing, node-link visualizations, and flow maps in cartog-
raphy [Wol13]. However, neither of these present a full survey or
comprehensive typology of geospatial network visualizations. The
application of visual analytics methods to geographic networks is
discussed by Rozenblat and Melançon [RM13], but their focus is
not on visualization methods as such, although they include an
overview of edge bundling methods. Jenny et al. [JSM*18] have
established design principles for flow networks. Finally, a variety
of geospatial network visualizations have been created by practi-
tioners, compiled in the online resource Visual Complexity (visual-
complexity.com).
Closest to our work, Hadlak et al. [HSS15] presented a survey
on the visualization of multi-faceted graph data, in which spatial
data is discussed as one possible facet of a multi-faceted graph.
The classes of techniques in our Composition dimension (Section
8) were informed by Hadlak et al.’s classification. However, Had-
lak et al. do not provide a deeper discussion specifically on geospa-
tial networks and their underlying visual representations. To the
best of our knowledge, ours is the most comprehensive survey on
the visualization of geospatial networks.
3. Scope and Definitions
Geospatial networks and their visualizations are used in many fields
such as information visualization, geovisualization, and graph
drawing and can include different types of data and terms. This
section aims to give an overview over the most common terms and
to define the scope of this survey.
3.1. Terminology for Data Types
Networks are considered synonymous to graphs in this survey. A
graph is formally defined as a pair G= (V,E), where Vis a set of
nodes (or vertices) and Eis a set of links (or edges), with each link
either being a set of two nodes (undirected graphs) or an ordered
set of two nodes (directed graphs). Both nodes and links can have
an arbitrary set of attributes AVand AEassociated with them.
In a spatial network, nodes are associated with inherent and
semantically meaningful spatial positions. For example, in brain
connectivity networks, nodes are distinct regions, and their posi-
tion information is essential for understanding brain activities. Net-
works with arbitrarily determined positions such as those generated
by network layout algorithms are not considered spatial networks.
Fixed node positions increase the difficulties when designing visu-
alizations for spatial network data.
Geospatial networks are a subgroup of spatial networks where
the node locations are of a geographic nature, i.e., these nodes
represent locations on the Earth or other planets. As locations on
the surface of approximately (but not exactly) spherical bodies,
geospatial locations have characteristics that differentiate them
from other spatial data, making the distinction between spatial and
geospatial essential. Besides precise geographic coordinates, loca-
tions can be defined more semantically and come with their own
set of challenges. For example, areas can be well-defined and non-
overlapping (e.g., Germany,France), roughly defined with fuzzy
boundaries (Sahara desert), nested (Scotland,UK), and overlap-
ping (Schengen Area,European Union). Geographic information
could point to multiple possible geographic positions of either cer-
tain value (Brest (France) or Brest (Belarus)) or uncertain value
(‘Mum’s house’ or ‘the forest’), or a geographic location could
point to entirely fictional places (‘Atlantis’). Some of these cases
introduce specific challenges to visualization. Reviewing the exist-
ing visualization and interaction techniques for geospatial network
data is the focus of our survey.
3.2. Related Concepts
There are several concepts adjacent to geospatial networks in that
they describe relationships or movements between different geo-
graphic locations. However, we do not consider all types of such
data geospatial networks.
Firstly, it is useful to think about these concepts in terms of link
continuity, which describes to what extent data about the trajectory
of each link is available. In some cases, like GPS tracking or air
traffic data, continuous physical routes between origins and desti-
nations are available. Such data are called trajectories. Trajecto-
ries have very high link continuity. Other data might be structured
such that full trajectories are not available but intermediate stops
between the origins and destinations are recorded, such as parcel
tracking data. Origin-Destination (or OD) data consists of only
origin-destination pairs, with no information on the trajectory be-
tween the two locations. OD data has the lowest link continuity.
Secondly, it is necessary to consider to what extent the locations
in the data (i.e., origins and destinations) represent network nodes.
If locations are essentially arbitrary, for example inthe case of ride-
sharing data where a trip can start and end at any given location (as
opposed to e.g., bus travel being limited to bus stops), each ‘node’
would only be connected to a single link. Here, further abstraction
in the form of aggregating individual origins and destinations into
areas would be necessary to form a meaningful network with nodes
that have more than one link each. Data types with higher link con-
tinuity can be abstracted to data types with lower link continuity;
submitted to COMPUTER GRAPHICS Forum (3/2021).

4S. Schöttler, Y. Yang, H. Pfister & B. Bach / Geospatial Network Visualization
for example, trajectories can be treated as OD data by ignoring the
information about trajectories between origins and destinations.
In summary, OD, trajectory, and other geospatial data can often
be interpreted as or abstracted to geospatial network data—which
is why our survey contains several techniques intended for these
data types—but not all OD or trajectory data should automatically
be considered geospatial network data.
4. Methodology and Design Space
Having defined the scope of our survey, this section details how
we gathered papers from scientific venues, removed irrelevant pa-
pers, explored different approaches for classifying and discussing
papers, and describes our final design space.
4.1. Collecting Papers
Contributions to geospatial network visualization have come from
many different fields including information visualization, graph
drawing, and cartography. To account for the diverse range of pub-
lication venues for potential visualizations of geospatial networks,
we followed a two-step approach. First, we looked at the proceed-
ings and collections of major venues where work on geospatial and
network visualization would naturally be published:
•IEEE VIS (InfoVis) (Accessed via [IHK*17])
•ACM Conference on Human Factors in Computing Systems
(CHI) (Accessed via [ACM19])
•IEEE/CGF EuroVis (Accessed via [Wil19])
•IEEE PacificVis / Asia-Pacific VIS (Accessed via [Pac19])
•Symposium on Graph Drawing (GD) (Accessed via [Spr19])
•ACM SIGSPATIAL conferences (Accessed via [SIG])
For each of these venues, we manually scanned the proceedings
and retrieved candidate papers based on their title. Papers were in-
cluded in this initial collection if their titles contained references
to geospatial networks and their visualization. In an effort to ob-
tain as many relevant papers as possible, we also included many
papers where the title only mentioned either networks or geovisu-
alization. We chose this approach because ‘geospatial networks’
are not a universally recognized concept, and many papers in our
collection use different terms. The resulting variety in terms being
used for similar concepts makes it problematic to identify relevant
papers based on title or keywords alone. This resulted in a set of
191 candidate papers.
In a second pass we manually selected the most relevant papers
out of the 191. To that end, we read the abstract and checked figures
in each paper. This reduced our collection to 41 papers. From these
41 papers, we then extracted keywords, which were used to perform
automated searches in the following online libraries which cover all
major visualization and geographic journals:
•ACM Digital Library
•IEEE Xplore
•Taylor & Francis Online (publisher of several geography-related
journals, e.g., Cartography and Geographic Information Science
&International Journal of Geographical Information Science)
•Google Scholar
The search terms (keywords) are a combination of different
terms to describe geospatial networks and terms used to describe
the visualization aspect. Terms were combined in a Boolean search
query as follows:
(geographical network(s) | geographic
network(s) | geospatial network(s) | spatial
network(s) | spatial interaction data |
origin-destination)
&
(visualization | visualisation | graph
drawing | flow map)
This yielded a large number of possible papers. Again, we man-
ually narrowed down the results of the search by examining at least
the abstract and figures in each of the retrieved papers. In addition,
a number of papers were discovered through following references
of some of the already retrieved papers as well as recommendations
from reviewers. This selection step yielded another 52 papers, rais-
ing the final number of papers to 95. A paper was included if both
of the following criteria were fulfilled:
•C1) A technique must be motivated by and designed for
geospatial networks and must visualize both geospatial and net-
work information. If a technique is not explicitly designed for
geospatial networks, its application must be demonstrated and
address a challenge in visualizing geospatial networks.
•C2) A paper must contain a novel and representative tech-
nique, rather than iterating or adapting existing techniques. For
example, we found many papers placing nodes at geographic po-
sitions and connecting them. Our survey does not list all these
papers but a manually chosen representative sample.
For example, we excluded an algorithm for clustering trajecto-
ries [AAFG18] or a technique for visualizing vessel movements
[WvdWvW09] because while trajectory or movement data can of-
ten be abstracted to network data, it was not in the context of these
techniques. Further, we excluded many edge and trail bundling
techniques and instead selected a sample intended to represent the
different possible types of edge bundling, which will allow for dis-
cussing implications for visualizing geospatial networks. In this
survey, the term edge bundling encompasses both edge and trail
bundling techniques, since the differences between the two cate-
gories are negligible in our context [LHT17b].
4.2. Creating a Design Space
Before arriving at our final design space, we went through several
iterations of taxonomies and typologies, each one informed by the
techniques themselves as well as alternative higher-level objectives.
For all iterations, a structured coding of our collection of papers
was performed by one of the authors, informing our assessment of
how useful each approach is. For the final design space, the full col-
lection was independently coded by two people and disagreements
resolved through discussion afterwards. The full, coded collection
is available on the website (geonetworks.github.io).
Version #1: Data-driven—Our first approach was to structure
techniques by the data types a given technique can be applied to,
e.g., directed or undirected networks, additional link attributes, or
dynamic geospatial networks. This approach is informative for de-
scribing how specific visualization challenges are addressed by the
submitted to COMPUTER GRAPHICS Forum (3/2021).

S. Schöttler, Y. Yang, H. Pfister & B. Bach / Geospatial Network Visualization 5
literature. However, this approach did not provide informative in-
sight into the visual characteristics of different techniques and help
discuss conceptual ideas behind techniques’ design. We decided to
discuss our grouping for #1 later in Section 10, complementing our
design space.
Version #2: Technique-driven—Our second approach involved
coding papers by grouping common visualization techniques for
networks such as node-link diagrams, adjacency matrices, flow
maps, etc. This is a common approach to classify techniques
across the visualization community as it often leads to a high-level
overview of major classes and ‘ideas’ of techniques. However, it
did not yield insightful results as most techniques (66%) used node-
link diagrams, most of which apply edge bundling. We were not
able to derive a meaningful discussion about how a specific chal-
lenge motivates a specific design and what problems it is solving.
Moreover, we felt we would fail to capture the richness of all the
different visualization approaches and to list meaningful directions
for future design.
Version #3: Challenge-driven—Our third intention was to cre-
ate a taxonomy around which problems and challenges a technique
addresses. We were hoping this would result in a practical resource
of solutions to common challenges. However, we found that such
an approach would suffer from three major drawbacks. First, a list
of challenges is necessarily incomplete if not derived from a sys-
tematic schema. In other words, without a systematic approach to
understand challenges in geospatial networks, any list would purely
capture the state-of-the art and our ‘taxonomy’ would be outdated
with the next technique proposing a solution to an unsolved (and
hence not appearing in our taxonomy) problem. Second, we would
potentially not be able to agree on the definition of a challenge. For
example, visual link clutter is a problem of node-link diagrams, but
not of geospatial networks themselves. For example, using adja-
cency matrices avoids this problem, rather than solving it. Lastly,
a taxonomy based on challenges would not help to discuss and un-
derstand design solutions, decisions, and to potentially inform new
designs. Together with Version #1, this informs our discussion of
challenges, and techniques addressing them, in Section 10.
Version #4: Representation-driven—We eventually decided to
code visualization techniques according to how they balance their
visual representations at the tension between i) explicitly showing
all possible information in a geospatial network, i.e., all links, all
nodes, all geographic information and places of interest, and ii)
managing visual clutter and information overload to provide for
efficient task-oriented visual representation. We found our design
space, dimensions, and classifications to best capture the trade-offs
required in designing geospatial network visualizations and to pro-
vide a conceptual framework perhaps similar to the design space
described by space-time cubes [BDA*17] or map-like visualiza-
tions [HHS20]. The complete rationale is given in Section 4.3.
4.3. Final Design Space
Our final design space consists of four dimensions and two sub-
dimensions (Figure 3), each representing an essential aspect of a
geospatial network visualization:
•D1: Geography Representation (GEO, Figure 3(a)) describes
how geographic information is visually represented. This dimen-
sion ranges from explicit to abstract visual encodings, where
(a) D1: Geography representation (GEO)
(b) D2.1: Node representation (NET:NODE)
(c) D2.2: Link representation (NET:LINK)
(d) D3: Composition (COMP)
(e) D4: Interactivity (INTERACT)
Figure 3: Dimensions and categories in our design space.
explicit implies a representation that uses a (cartographic) map
projection. Such mapped (explicit) visual encodings follow the
underlying principle of geographic maps: mapping geospatial lo-
cations, which are naturally three-dimensional, onto a 2D plane
by using a map projection. Alternatively, locations may be dis-
played on a 3D globe. Abstract encodings use visual encodings
not based on map projections. These encodings use alternative
submitted to COMPUTER GRAPHICS Forum (3/2021).

6S. Schöttler, Y. Yang, H. Pfister & B. Bach / Geospatial Network Visualization
ways of encoding geographic information, thus potentially leav-
ing more visual space and visualization options (position, visual
marks, etc.) for encoding network and other information in the
visualization. Between mapped and abstract techniques, we can
find techniques that distort geography (distorted).
•D2: Network Representation (NET, Figure 3(b+c)) describes
how topological information of the network is represented. Like
GEO, this dimension describes visualization techniques along a
spectrum ranging from explicit to abstract. An explicit encoding
uses a one-to-one mapping, where each node and link is encoded
as a separate visual element. This allows for precise topological
tasks such as assessing if two nodes are connected. Aggregated
encodings use one-to-many mappings, where multiple nodes or
links are represented as one visual element. An abstract encod-
ing uses a more complex mapping of topological data to visual
elements. Decoding the visualization to extract low-level topo-
logical information is often not possible, but an abstract network
encoding can make more visual space and encodings available
for the geographic aspect of the data.
As we found techniques that abstract only one of both nodes or
links, we classify techniques separately for nodes and links.
•D2.1: Node representation (NODE, Figure 3(b)) describes how
nodes are represented: explicit,aggregated, or abstract.
•D2.2: Link representation (LINK, Figure 3(c)) describes how
links are represented: explicit,aggregated, or abstract.
The dimensions GEO and NET can be seen as two sides of the
same coin: making decisions about the visual representation of net-
work topology implies a decision on how to represent geography.
For example, a visualization might choose to explicitly represent
geography and maintain spatial distances between locations. This
will most necessarily result in issues with node overlap if nodes are
placed at similar or nearby locations. Or links overlap for the same
reason and can stretch far over the geographical representation. On
the other hand, a designer could choose to abstract geographic in-
formation, e.g., by distorting geographical distances to provide for
better perception and understanding of a network’s topology. Dif-
ferent visualization techniques propose different solutions to over-
come this tension and abstract information in the network as we
discuss in Sections 5to 7.
Besides GEO and NET, we include two further dimensions.
•D3: Composition (COMP, Figure 3(d)) describes how network
and geographical information are integrated visually. Inspired by
Hadlak et al.’s composition mechanisms for multi-faceted net-
works [HSS15], this dimension runs from a loose integration
(e.g., juxtaposition) to a strong integration.
•D4: Interactivity (INTERACT, Figure 3(e)) describes to what
extent a technique requires user interaction for exploring and
connecting geography and network data of a geospatial network
or whether a technique is an interaction technique in its own right
(Interaction Only).
With this dimension-driven approach, we can capture the rich-
ness as well as some of the design decisions in existing techniques,
while at the same time providing a design space to locate and com-
pare existing techniques as well as inform discussions about miss-
ing approaches. Each dimension classifies techniques according to
how much of that information (geography, nodes, links) they show
explicitly, and how much of that information they abstract and ag-
gregate. Along each dimension, techniques and designs are roughly
grouped into categories, although transitions between these cate-
gories are fluent. The following Sections 5to 9detail each of these
dimensions and discuss the types of representations, compositions,
and interactions we found. A discussion on the limitations of this
approach is provided in Section 11.
5. D1: Geography Representation (GEO)
The GEO dimension describes how geographic information is rep-
resented visually. Geographic information includes geographic lo-
cations (or node locations) as well as more general information
related to spatial distances, regions, landmarks, relations between
these locations (hierarchical, distances, etc.), and any additional ge-
ographic information important for the visualization. For GEO, we
define three major categories along a continuous explicit—abstract
spectrum (Figure 3(a)): mapped,distorted, and abstract. Note that
these ‘categories’ are not discrete sets, but rather steps along a con-
tinuous spectrum.
5.1. D1—GEO: Mapped
Mapped techniques are the most explicit geographic
representations. In mapped techniques, geospatial lo-
cations are visually represented by positioning visual
elements using a geographic map projection. Any map
projection introduces distortion as it is impossible to flatten a three-
dimensional surface into a two-dimensional map without any dis-
tortion [Sny87, p. 3]. However, different map projections preserve
different features of the geography, e.g., angles, areas or some dis-
tances, and as such the choice of projection is always a trade-off
between different kinds of distortion. In addition, the mapped cat-
egory includes three-dimensional globes. We chose not to differ-
entiate between two-dimensional and three-dimensional represen-
tations as separate categories in this design space because the dis-
tinction is not about 2D or 3D but about a consistent presentation of
a geographical space. Distortion, as introduced in the next section,
distorts a geographic space based on the network data and thus in-
troduces a data-driven distortion. For example, 3D globe represen-
tations can be distorted in the same way as 2D maps, based on the
network’s topology [ASB07]. The majority of the surveyed tech-
niques use mapped representations (74%).
(a) (b)
Figure 4: Examples for GEO–Mapped: (a) A globe in
VR [YDJ*19], (b) A flow map superimposed on a map; it
uses an automatic layout based on spiral trees [BSV11a;BSV11b]
(Reprinted by permission from Springer Nature: Springer. Lecture
Notes in Computer Science. “Angle-Restricted Steiner Arbores-
cences for Flow Map Layout”, Buchin, K., Speckmann, B., and
Verbeek, K. © 2011).
submitted to COMPUTER GRAPHICS Forum (3/2021).

S. Schöttler, Y. Yang, H. Pfister & B. Bach / Geospatial Network Visualization 7
Three-dimensional (3D) globe representations offer the most
precise geographic information with respect to geographic area
sizes and distances. 3D globes, either displayed on screens or in
virtual reality (VR) [YDJ*19] (Figure 4(a)), have been used to dis-
play node-link diagrams in various forms, e.g., using straight lines
on a globe [CEH96], arcs on a globe [MHCF96;YDJ*19;KB16],
edge bundling around the globe [LBA10b;ZZLL18], or flow maps
spanning a globe [DSD14b]. 3D globes preserve global distances
and sizes of areas. Also, link crossings can be reduced since links
can be drawn along their shortest path around the globe, naturally
routing links around each other. The main shortcoming of globes is
that, unless the entire network is located on one half of the globe,
interaction is required for full exploration.
However, most mapped techniques use 2D maps, despite the
different types of distortions introduced by different map projec-
tions (such as Mercator or others) [Bat09]. Still, 2D maps are
highly usable on 2D screens and in print media. Many visualiza-
tions show geographic details such as roads [AAFW17] and coun-
try borders [ITH16]. Other techniques reduce geographic detail to
country shapes [BSV11a] (Figure 4(b)) or remove any geographic
detail except locating nodes at their respective geographic position
on the screen [GHNS11;BW98].
(a)
(b) (c)
Figure 5: Examples for GEO–Distorted: (a) OD map [WDS10]
(reprinted from [KSDW13]), (b) A deformed map, visualizing travel
times by train in the UK [BDD*16] (reprinted from [Bou17]),
(c) An automatically created and labelled metro map [NW11]
(reprinted from [Nöl09]).
The major advantage of mapped representations is that they sup-
port tasks related to a) purely geographic information such as
Which regions are close? How far are these locations apart? Which
country is this? Is there a mountain? and b) geographic informa-
tion about the network topology (if topology information is pro-
vided properly by the network representation): Is this region well-
connected? How far apart are these two nodes? Which one is the
longest link in my network? A second advantage of mapped tech-
niques is that people are familiar with geographic data being dis-
played on maps. Maps have been shown to have cognitive benefits
when interpreting geographic data [HHS20].
The drawback of mapped techniques is that they cause prob-
lems when nodes are close or at the same position, or when links
span large distances. Depending on zoom level, nodes will often be
displayed off screen while at the same time being highly related to
the currently visible nodes. Abstracting the geographic information
through distortion, aggregation, or removing information can offer
some solutions. Another issue present in all mapped node-link di-
agrams is the visual dominance of long links. A connection across
continents is not necessarily more important than a local link (of-
ten quite the opposite), yet takes up much more space simply due to
the geographic context, potentially covering shorter links as a side
effect. This is further aggravated through the distorted distances
caused by most map projections. Finally, mapped representations
result in positional variables of nodes not being available for other
information such as connectivity or node type.
5.2. D1—GEO: Distorted
Distorted describes techniques where geospatial loca-
tions are displaced with respect to their original posi-
tion in the initial map projection—both 2D and 3D. As
discussed in Section 5.1, any map projection introduces
some distortion as a consequence of projecting 3D space onto a 2D
plane, but the Distorted category deals with distortions based on the
network data, either directly through data-driven algorithmic dis-
tortions, or indirectly through interaction techniques that let users
distort the visualization to explore the network. Generally, distort-
ing the geography is used as a way to show the network topol-
ogy more clearly and avoid difficulties such as described in the last
paragraph of the previous section.
For non-spatial networks, laying out nodes based on their con-
nectivity is common practice. For example, force-directed layouts
place nodes with strong connectivity (many connections and many
common neighbors) closer together. For geospatial networks, there
is an inherent conflict between using node positions to represent ge-
ographic locations and using them to expose the topology. Distort-
ing the underlying map or displacing individual nodes has been ex-
plored as a compromise to address this conflict. Among the 20% of
techniques that use distortion, we identified five main approaches,
including both continuous and discontinuous types of distortion:
First, there are techniques that use continuous distortion to show
alternative measures of distance instead of the geographic dis-
tance between two nodes [ASB07;BDD*16] (Figure 5(b)). Alter-
native measures of distance in this context could be for example
travel times or dissimilarity measures.
Second, tile maps transform geographic regions into a grid of
identical tiles, a form of discontinuous distortion. The tiles usually
submitted to COMPUTER GRAPHICS Forum (3/2021).

8S. Schöttler, Y. Yang, H. Pfister & B. Bach / Geospatial Network Visualization
cannot be placed at their original geographic locations [MH17].
For example, OD maps [WDS10] use nested tile maps to represent
geospatial networks. As illustrated in Figure 5(a), each cell con-
tains a small version of the larger map. The color of each small
cell indicates the flow volume into that cell from the larger cell it is
nested into.
Third, map insets are a form of discontinuous distortion. Map
insets either contain smaller (undistorted) sub-maps at different
scales from the main map [BKA*16;OHN*18] (Figure 15), or in-
dividual nodes [GRE11]. Insets scale parts of the map up or down,
or move locations away from their original position, which we con-
sider a form of distortion since it results in distances, areas and
angles no longer being consistent across the map.
Fourth, there are representations where node positions are com-
puted as a trade-off between showing the true geospatial location
and clearly showing the network topology, usually using continu-
ous distortions. Essentially, nodes are shifted to increase the legibil-
ity of the network representation. An example of this is centrality-
based scaling [MG06], where the underlying geography is distorted
such that dense areas in the network are enlarged compared to
sparser areas, while preserving link orientation as much as pos-
sible. A similar method specifically for road networks is proposed
by Haunert and Sering [HS11]. A notable application of this type of
distortion are metro map layouts [e.g., HMdN06;WC11;BBDZ08;
vDL19]) (Figure 5(c)), a classic application of graph drawing. The
common place of such transit maps is that users do not need to
know the precise geography of the transport network—it is more
important that they can clearly see how different lines connect so
that they can plan their route accordingly. As such, a map that dis-
torts the geography to the extent necessary to create an easily legi-
ble network map is an ideal trade-off.
Lastly, continuous as well as discontinuous distortion can be
used as an interaction technique, scaling different parts of the
network up and down based on user interaction. A fisheye lens
[BMS93] is a classic implementation of this. For metro maps, a
custom scaling method specifically for this purpose has been pro-
posed [WC11]. Interaction techniques are discussed separately in
Section 9.3.
5.3. D1—GEO: Abstract
An abstract geography representation encodes geo-
graphic information without the use of map projec-
tions. This usually includes non-spatial variables such
as color or shape (Figure 6) and can be seen as the
exact opposite of explicit geographic representations. Often, geo-
graphic locations are aggregated into coarser groups. Note that a
visualization can still use position to place visual marks on the pic-
ture plane, but in abstract geographic representations there is no
natural mapping between an element’s position on the screen and
its geospatial location. Displaying locations along a line could be
considered a projection from 3D to 1D, but it is not a map pro-
jection, which is defined as a projection onto a 2D plane [Sny87,
p. 3].
Abstract representations allow for a great variety of designs and,
we believe, offer many unexplored solutions as we only found five
papers with abstract geographic encodings (5%). Two papers use
(a) (b)
(c)
Figure 6: Examples for GEO–Abstract: (a) ‘Kriskograms’, the lo-
cations are projected to positions on a one-dimensional straight
line [XC09] (reprinted by permission of the publisher, Taylor &
Francis Ltd, www.tandfonline.com), (b) Global international mi-
gration flows shown in a chord diagram, the locations are grouped
and color-coded [AS14] (Figure courtesy of Federal Institute for
Population Research (BiB), Germany), (c) ‘OntoTrix’, geographic
nodes are grouped (‘City’ node) [BPLL11].
circular chord diagrams [Hen13;AS14] (Figure 6(b)) in which
geographic information is encoded as groups of nodes along the
circle (technically along a single spatial dimension, i.e., the cir-
cle’s circumference), using color as redundant encoding for geo-
graphic regions. A third technique uses a variation of arc diagrams
termed Kriskograms [XC09] (Figure 6(a)), which orders nodes on a
horizontal 1-dimensional straight line. Our specific example orders
nodes according to their position from west to east but other encod-
ings, e.g., grouped by country as in the chord diagrams are easily
imaginable. In both examples, geospatial locations can be approx-
imated with greater or less detail, e.g., locations on the Northern
and Southern as well as Eastern and Western Hemisphere. A fourth
technique is an adjacency matrix, in which nodes can be grouped
by geographic location or region [BPLL11] (Figure 6(c)) or geo-
graphic regions being integrated into the network topology as in-
dividual nodes, e.g., a node for every location having links to the
nodes related to these locations. Further examples of abstract en-
codings could include coloring nodes in a force-directed node-link
layout based on their geographic location, e.g., country.
The GEO dimension represents a continuous spectrum and pro-
vides the designer with many choices and opportunities. Decisions
depend on the level of precision required for the geospatial aspect
submitted to COMPUTER GRAPHICS Forum (3/2021).

S. Schöttler, Y. Yang, H. Pfister & B. Bach / Geospatial Network Visualization 9
of the data. For example, if a task requires geographic fidelity (e.g.,
estimating spatial distribution, densities, and distances of nodes),
then mapped representations should be naturally considered. If,
however, a task can neglect certain geographic fidelity (e.g., esti-
mating number of nodes in a given region), geographic information
could be abstracted or distorted to provide space for visualizing in-
formation relevant to the task (e.g., distorting space to remove over-
lap between nodes to support estimating the number of nodes in a
given region).
6. D2.1: Node Representation (NODE)
The NODE dimension is the first subdimension of NET and de-
scribes the visual representation of the nodes of the network along
an explicit—abstract spectrum. In a geospatial network, nodes rep-
resent locations or geolocated entities, and they are related to each
other by links.
The most explicit representation is one where each node is in-
dividually visually represented, either through displaying a visual
element such as a dot, or by otherwise clearly indicating its po-
sition. For example, node symbols are often omitted in node-link
diagrams, but the start and end points of links clearly indicate node
positions (e.g., Figure 11(a)). The most abstract representation is
one where individual nodes are not visually indicated at all, making
it hard to reconstruct the overall network topology. In between these
two extremes, we find that nodes can be aggregated into groups—
each node is assigned to a specific group, but it is not displayed as
a separate element. The node representation is entirely independent
from the link representation, and any of the three NODE classes can
be combined with any of the three LINK classes.
6.1. D2.1—NODE: Explicit
With 87% of the surveyed techniques, the majority of
techniques show nodes explicitly, i.e., show each indi-
vidual node in the network. In any of these represen-
tations, nodes can be shown explicitly as points (node-
link diagrams, arc diagrams) or rows and columns (adjacency ma-
trix) (Figure 7(b)). Sometimes, especially in techniques using edge
bundling, explicit visual markers for nodes are omitted but node
locations are still clearly identifiable from the endings of links.
Most techniques with explicit node representations are variations
of node-link diagrams, differentiated only by the types of links they
use (such as straight lines, arcs, bundled edges, etc.). These types of
visualizations are typically superimposed on a mapped geography
representation.
However, explicit node representations have also been used in
combinations with both explicit and distorted geography. For ex-
ample, Necklace maps [SV10] (Figure 7(a)) display nodes on a cir-
cle surrounding the relevant part of the map. Nodes are placed on
the circle according to their geospatial position, resulting in a dis-
torted display of the geography. A similar method is proposed by
Stephen and Jenny [SJ17], where nodes that are out of view are
laid out around a target area in a circular arrangement and con-
nected to nodes in the target area (Figure 14(b)). The distorted
globe by Alper et al. [ASB07] (Figure 16(b)) and deformed map
by Bouts et al. [BDD*16] (Figure 5(b)) use explicit nodes on a
globe/map, combined with an abstract link representation.
(a) (b)
Figure 7: Examples for NODE–Explicit: (a) A ‘Necklace map’; the
space in the center can be used for displaying links between the
nodes on the outer circle [SV10;SV15] (reprinted from [SV15]),
(b) An Origin-Destination (OD) Matrix.
Combined with an abstract geography representation, we have
found techniques that display nodes in a circular, spatially ordered
layout, or on a single spatially ordered axis, with the network topol-
ogy shown as a chord diagram [AS14;Hen13] (Figures 6(b) &
16(a)) or arc diagram [XC09] (Figure 6(a)).
6.2. D2.1—NODE: Aggregated
Aggregated node representations group individual
nodes into metanodes and show only these metanodes
explicitly. Aggregation can happen through grouping
nodes in predetermined areas [Guo09], or more flexi-
ble partitions either through user interaction, e.g., specifying ge-
ographic regions [vdEvW14] (Figure 8(a)), or by algorithmically
identifying dense clusters and grouping nodes by these clusters
[LBW17b] (Figure 8(b)). The OD map [WDS10] (Figure 5(a)) uses
a regular spatial grid or tiling, overlaid onto a 2D map and groups
nodes within each grid cell. The volume of flows (link weight) be-
tween each pair of cells is then further encoded inside the cell using
a nested map and color.
Aggregating nodes reduces the number of visual elements dis-
played in the visualization. This can automatically reduce the num-
ber of links as they can be aggregated into metalinks, i.e., links
between metanodes. In reducing detail, aggregation can highlight
higher-level patterns and allows for showing summary statistics and
data for each metanode (Figure 8(a)-right).
6.3. D2.1—NODE: Abstract
While aggregated representations still visualize net-
work topology to some degree, abstract node represen-
tations do not show any visual marks that are identifi-
able as individual nodes or metanodes but rather com-
municate approximate locations and areas with fuzzy boundaries
where individual nodes are situated.
We found five examples of abstract node representations (5%),
all of which were created for flow data. Guo and Zhu use a flow-
based density estimation method to extract patterns from the flow
data [GZ14] and then show these patterns as flows on a map. The
same authors propose a second, similar method, where flows are
clustered based on similarity, and each cluster displayed as a single
submitted to COMPUTER GRAPHICS Forum (3/2021).

10 S. Schöttler, Y. Yang, H. Pfister & B. Bach / Geospatial Network Visualization
representative flow [ZG14] (Figure 9). Using these abstract tech-
niques, the geographic representation can show location-specific
information (attributes in color) such as bandwidth or unemploy-
ment, instead of overloading the visualization with explicit or ag-
gregated nodes. This information can be visually correlated with
the flows while providing detailed geographic information. Sim-
ilarly, Kim et al. [KJW*18] propose a combination of heatmaps,
‘field line’ maps (Figure 12(a)), and arrows superimposed on a map
to visualize flows, whereas Yao et al. [YWZ*19] use hexagonal pat-
tern maps to indicate directions (Figure 12(b)).
As these techniques show, abstracting nodes to fuzzy areas or
glyphs can reveal overall patterns of how different areas are con-
nected, especially for very large and dense networks. However,
since any abstraction selectively hides some details of the data, the
quality of the final visualization heavily relies on choosing methods
appropriate to the data and task.
In summary, the NODE dimension, similar to GEO, represents
a continuous spectrum that provides a range of possible design
choices: showing nodes explicitly and with full detail, aggregating
nodes into clusters and other meaningful groups, and eventually
fully abstracting the visual node representation, focusing only on
connectivity. The tasks supported by this range depend on the re-
quired detail about nodes; the less individual nodes are important,
the more nodes can be abstracted and the visualization be designed
for tasks focusing on, e.g., density, geographic landmarks, etc.
7. D2.2: Link Representation (LINK)
Link representation is the second sub-dimension of the network
representation dimension (NET). Similarly to NODE,LINK ranges
from explicit to abstract, and we classify techniques into one of the
three classes explicit,aggregated, and abstract.
(a)
(b)
Figure 8: Examples for NODE–Aggregated: (a) The user can inter-
actively select regions to aggregate nodes [vdEvW14], (b) Module-
based visualization: Nodes are automatically aggregated into clus-
ters [LBW17b] (Reprinted by permission from Springer Nature:
Springer. Journal of Visualization 20, 205–215. “Module-based vi-
sualization of large-scale graph network data”, Li, C., Baciu, G.,
and Wang, Y. © 2017).
7.1. D2.2—LINK: Explicit
The largest group of techniques in this class are tech-
niques based on node-link diagrams, which are largely
differentiated by the visual representation of their links.
A flow map is essentially a special case of node-link
diagram, in which all links are weighted and directed, represented
by arrows of varying widths. The simplest form of flow map uses
straight lines or arrows to connect nodes, superimposed on a reg-
ular map. Particularly early computer-generated visualizations use
this approach [e.g., Tob87;BEW95]. However, straight lines create
cluttered maps even with comparably small data. As a consequence,
a multitude of techniques have been developed to reduce clutter in
node-link diagrams and flow maps, of which many are based on
aggregation and as such discussed in the following section.
In smaller, less dense networks, clutter is less of an issue.
Yet, links may overlap, or multiple links may connect the same
nodes. To address this, several techniques have been developed
to ensure links are clearly visible as separate lines. This can
be achieved by bending them, for example using Bézier curves
[BW98], which can be further spread out using angle constraints
where the links connect to the nodes [BST00]. In three-dimensional
visualizations, lifted arcs may reduce overlap compared to a two-
dimensional display, particularly when the user can interactively
navigate the visualization [YDJ*19;VFAA17] (Figures 4(a) and
10(b)). Kaya et al. [KB16] introduce a technique to spread out links
around a globe.
In addition to reducing overlap, another challenge in node-link
diagrams is the visualization of link attributes, which can be as sim-
ple as direction or weight, but also take more complex forms. When
using arcs, the arc height can be used to encode data attributes such
as link weight or distance [VFAA17] (Figure 10(b)). For networks
with additional link attributes more complex than weights or direc-
tions, different modifications to the links have been proposed, such
as animated link textures [RAB*18], where shape, size, color, and
direction of particles moving along the links can be used to encode
attributes. A static variant are patterned links [CKS*16].
In a flow map, each link would typically be represented by a line
of uniform thickness, determined by its weight. An alternative to
this are tapered links. In addition, there are different shapes and
sizes of arrowheads that can be used [KLSC12].
Dynamic geospatial networks are networks that change over
time. Changes can be limited to certain attributes, e.g., link weights,
Figure 9: Example for NODE–Abstract: No nodes are shown in this
technique by Zhu and Guo, the displayed ‘links’ have been obtained
by clustering together links of a very dense network [GZ14].
submitted to COMPUTER GRAPHICS Forum (3/2021).

S. Schöttler, Y. Yang, H. Pfister & B. Bach / Geospatial Network Visualization 11
(a) (b)
(c)
Figure 10: Examples for LINK–Explicit: (a) An interactive
flowmap created with flowmap.blue [Boy20], (b) A directed node-
link flow map with 3D arcs [VFAA17], (c) ‘MapTrix’, a matrix con-
nected to two maps [YDGM17].
but can also affect the entire network, including new or deleted
nodes or links. Dynamic networks can be visualized through an-
imating node-link diagrams or displaying them as small multi-
ples [BBL12]. Flowstrates is an alternative for data that can be
represented as a directed, bipartite graph. The flows are displayed
using two maps, one for origins and one for destinations. The two
maps are linked through a heatmap showing the flows between the
linked locations over time. Figure 13 shows refugee flows as an
example, which change over time, and often go from one region of
the world to another [BBBL11]. Displaying the network in a space-
time cube is another method [KW05].
Node-link diagrams have been adapted to visualize uncertainty
in the underlying graph data. Schulz et al. [SNG*17] introduce a
technique that first decomposes the uncertain graph into its possi-
ble instances, then creates a visualization from these. Von Landes-
berger et al. [LBW17a] make further suggestions for visualizing
uncertainty in geospatial network data.
A variety of interaction and navigation techniques have been de-
veloped for node-link diagrams. These are discussed in Section 9.3.
A different form of explicit link representation is used by
matrix-based techniques, where each node is represented by a
row and column in the matrix, representing outgoing and incom-
ing flows to other nodes. Representing the network as a ma-
trix requires the geospatial aspect to be displayed separately.
This can be done on a juxtaposed map where color-coding re-
lates the locations to the matrix [Guo07]. For dense, directed,
and weighted geospatial networks (often called many-to-many flow
(a)
(b) (c)
Figure 11: Examples for LINK–Aggregated: (a) Force-directed
edge bundling [HvW09b], (b) Edge bundling and splatting on a
3D globe [LBA10a], (c) A one-to-many flow map [Sun19].
data), Yang et al. [YDGM17] introduce MapTrix, a technique that
shows a matrix with a 45◦rotation, rows and columns connected
to their associated location on two juxtaposed maps (Figure 10(c)).
As such, MapTrix employs the same basic concept as Flowstrates:
inserting a graph representation in between two maps showing ori-
gins and destinations.
7.2. D2.2—LINK: Aggregated
Links are considered aggregated if they cannot be in-
dividually identified anymore. Aggregated links can be
a good solution to overlap and clutter on node-link di-
agrams, and as a result many (28%) of our surveyed
techniques fall into this category.
When nodes are aggregated, this nearly always leads to aggre-
gated or abstract link representations because most visual represen-
tations remove, hide, or combine links between aggregated nodes.
Examples are Van den Elzen and Van Wijk’s technique for multi-
variate network exploration [vdEvW14] or Li et al.’s module-based
visualization [LBW17b], both shown in Figure 8. However, the ma-
jority of aggregated link techniques we found use explicit node rep-
resentations, aggregating the links only. Note that nodes can be ag-
gregated while links are explicit: in such a case, individual links
would be shown between clusters of nodes. However, we did not
find such a technique applied to geospatial networks.
Edge bundling, for which numerous algorithms have been pro-
posed, is a technique that bends links to form bundles [e.g.,
HvW09b;LHT17a;PHT15;PLCP12;GHNS11;CZQ*08] (Fig-
ure 11(a)). The level of aggregation depends on the algorithm
and parameters that are chosen — if the bundling is very ‘loose’,
with individual links clearly visible, the representation may in fact
still be considered explicit. However, most edge bundling tech-
niques generate clearly aggregated results, for example the 3D edge
bundling algorithm by Lambert et al. [LBA10a] (Figure 11(b)). A
more complete overview of edge bundling techniques is provided
submitted to COMPUTER GRAPHICS Forum (3/2021).

12 S. Schöttler, Y. Yang, H. Pfister & B. Bach / Geospatial Network Visualization
by Zhou et al. [ZPYQ13] and Lhuillier et al. [LHT17b]. Edge rout-
ing is an alternative approach that defines fixed routes that links
must follow, effectively also bundling edges [BS15].
Flow maps, i.e., node-link diagrams with directed, weighted
links, suffer from clutter problems even for relatively small data
sets. Link weights are represented by line thickness, and arrow
heads are added to each link to show its direction, resulting in
thicker links easily covering up nodes and other links. This can be
somewhat mitigated by changing the styling of the links and draw-
ing thinner links on top of thicker ones [JSM*18], but nonetheless,
straight-line layouts are severely limited. To mitigate this, differ-
ent types of bundled flow map layout algorithms have been pro-
posed: a layout based on combining edge routing and spline in-
terpolation [DLY*05], a layout based on spiral trees [BSV11a]
(Figure 4(b)), stub bundling [NB13], and a force-directed lay-
out [JSM*17]. To create a layout similar to that of the original
flow maps by Charles Minard [reproduced in Ren18], Sun proposes
a layout that routes links through oceans where possible [Sun19]
(Figure 11(c)). Additionally, several ‘hybrid’ techniques between
edge bundling and bundled flow maps exist. These techniques take
account of the link direction when bundling spatially proximal
links, for example, divided edge bundling which bundles flows
in different directions separately [SHH11] or Graser et al.’s edge
bundling-based flow map technique [GSRB19]. There are also flow
map techniques developed for three-dimensional globes [DSD14b].
7.3. D2.2—LINK: Abstract
Abstract link representations are those that do not dis-
play each link as an explicit visual element. In general,
we observe that abstract link representations are often
used to visualize a specific aspect of the network data
rather than provide an overview of the network topology in general.
Several techniques focus on visualizing the directional aspect of
origin-destination network data. To visualize the direction of flows,
Andrienko et al. [AAFW17] propose a technique where node lo-
cations are marked by glyphs, which show the strength of flows in
different directions over time. Pattern maps [YWZ*19] visualize
flows on a hexagonal grid which is superimposed on the geogra-
phy. Each hexagon indicates the magnitude of the flow on its inside
(grey), and the distance travelled on its boundary (red/orange) (Fig-
ure 12(b)). Kim et al. [KJW*18] propose a technique that generates
‘field lines’ to show flow strength and direction (Figure 12(a)).
Other techniques focus on showing the ‘connectedness’ or
similarity of different nodes. The distorted globe technique by
Alper et al. [ASB07] shows the node locations on a globe, distort-
ing the globe such that nodes are closer together the more closely
linked they are, without showing links as arcs or similar (Fig-
ure 16(b)). A similar technique on a flat map has been introduced
by Bouts et al. [BDD*16] (Figure 5(b)).
Finally, several techniques exist that extract high level patterns
from the network data. Andrienko et al. [AA11] developed a tech-
nique that at first glance, seems to show a regular flow map, yet
not only makes use of nodes aggregated into areas, but also of
partitioned trajectories such that flows are only drawn between
neighboring areas. The previously discussed techniques by Guo
and Zhu [GZ14;ZG14] follow this pattern as well (Figure 9).
(a) (b)
Figure 12: Examples for LINK–Abstract: (a) Showing patterns
using ‘field lines’ [KJW*18], (b) Detail and legend of a ‘pattern
map’ [YWZ*19].
In summary, the LINK dimension describes the spectrum of in-
formation shown for links. Explicit link representations support
tasks about individual links (e.g., path following, link type, link
direction), while aggregating links supports general connectivity
tasks (e.g., high-level network structure, connectivity of regions).
Abstracting links can help showing direction and strength of links
without occluding any geographic features. This is helpful where
geographic features or nodes are important, where nodes are very
closely placed on a map so that space for visualizing links is scarce,
or where a specific network metric (e.g., density per region) is im-
portant and can best be shown by showing meta-information about
links and thus choosing a more abstract representation rather than
displaying individual links.
8. D3: Composition (COMP)
This dimension describes how geographic (GEO) and network
topology information (NET) are composed into a single visualiza-
tion to provide for tasks that potentially involve information from
both: Which region has the most nodes? Which regions are most
connected to each other? What is the location of a particular node?
For combining general views in visualization, Javed and
Elmqvist [JE12] propose a taxonomy consisting of superimposi-
tion,overloading,juxtaposition,integration, and nesting. In their
survey on multi-faceted graphs, Hadlak et al. [HSS15] use similar
categories: juxtaposition,superimposition, and nesting.
Based on our collection of techniques and inspired by both Javed
and Elmqvist as well as Hadlak et al. we define four types in which
geographical and network information are combined in geospa-
tial network visualizations, ranging from a loose integration (jux-
taposed) via superimposed and nested to a strong integration (inte-
grated).
8.1. D3—COMP: Juxtaposed
In a juxtaposition, network representation and ge-
ography representation are placed side-by-side on
the screen as commonly done in Coordinated Mul-
tiple Views (CMV) environments. In a CMV sys-
tem [Rob07;WWK00], views are meant to show complemen-
tary information and each view can act as selection tool to fil-
ter or select information for other views. Brushing and linking
are common interactions for these systems. Typical examples for
submitted to COMPUTER GRAPHICS Forum (3/2021).

S. Schöttler, Y. Yang, H. Pfister & B. Bach / Geospatial Network Visualization 13
Figure 13: Example for COMP–Juxtaposed: ‘Flow-
strates’ [BBBL11].
juxtaposed visualizations for geographic networks include the OD
Wheel [LWLY15] and Ibarra et al.’s juxtaposition of a map and
an arc diagram [ITH16]. Techniques that use multiple represen-
tations of either geography or network, e.g., a 3D globe and a
2D map [CEH96]) or multiple perspectives on network topol-
ogy [Guo09] do not count as juxtaposed in our survey.
To better relate elements across different views of network
topology and geography, some techniques draw leader lines be-
tween locations and nodes, e.g., between regions on a map and
rows and columns (the network’s nodes) of an adjacency ma-
trix [YDGM17] (Figure 10(c)). Another example of juxtaposition
is Boyandin et al.’s Flowstrates [BBBL11] (Figure 13) which com-
bines two geographic maps with a third visualization in between
the two maps, showing changes in link weight over time. Each of
the timelines in the central visualization is linked to the respective
regions in both maps to provide for a network representation.
Juxtaposition strongly reduces visual clutter and allows for com-
bining precise and mapped geographic representations with any
type of network representation (node-link, matrix, etc.), at the same
time allowing for explicit representation of nodes and links. How-
ever, as pointed out by Javed and Elmqvist [JE12], users need to
build a mental mapping between the two representations. This can
require extra mental effort, especially when multiple objects are
targeted. Elements need to be connected across visualizations us-
ing visual elements such as lines or interaction such as brushing
and linking. Also, the visual space is divided, so each representa-
tion has less visual space for presentation.
8.2. D3—COMP: Superimposed
In superimposed compositions, the network and ge-
ography representation are visually overlaid, in most
cases the network on top of the geography. Superimpo-
sition can happen in that geography is seen as a refer-
ence for placing nodes at their precise geographic positions (Fig-
ure 14(a)) or as approximation for node positions (Figure 14(b)).
It also does not matter whether geography is mapped or distorted,
or abstract. As discussed in Section 5.3, in abstract geographies,
geography can still be shown along a single spatial dimension
while links are overlaid (Figure 6(a)). Still, the most commonly
found versions of superimposition (55%) overlay explicit nodes
on mapped geographies [DLY*05;BSV11a;JSM*18;Sun19;
YDJ*19] (Figure 4(b),11(c) &14(a)). Some techniques also su-
perimpose abstract node representations on a mapped geography
representation [AAFW17;YWZ*19] (Figure 12(b)).
(a) (b)
Figure 14: Examples for COMP–Superimposed: (a) A flow map us-
ing curved arrows [JSM*18] (reprinted by permission of the pub-
lisher, Taylor & Francis Ltd, www.tandfonline.com), (b) Nodes laid
out around a map [SJ17].
An interesting variation of this approach is In Situ Explo-
ration [HSS11], where the user can draw rectangular ‘portals’ onto
the map and select a type of visualization, e.g., a matrix. The ‘por-
tal’ inset then displays a visualization of the data of the region
which it covers on the underlying map.
Superimposition can often be easier to interpret and creates less
cognitive load for the viewer. However, superimposition has visual
elements of both geography and network in the same visual space,
which may easily produce visual clutter and might make it difficult
to visualize additional attributes.
8.3. D3—COMP: Nested
In a nested representation, one of network or geography
representation is nested inside the visual elements of
the respective other representation (geography or net-
work). One example of a nested representation is the
OD Map by Wood et al. [WDS10]. An OD Map divides a geo-
graphic map into cells in a regular grid. Then, inside each cell, i.e.,
a confined geographic space, a small representation of the entire ge-
ographic space is nested (Figure 5(a)). Geographic regions on each
of the small maps can now be used to encode information about
the links between the region represented by the cell and each re-
gion on the miniature map. Hadlak et al. and Schulz et al. propose
techniques that nest point-based tree representations of hierarchical
data into geographic regions on a map [HTSS10;SHS11].
Another nested technique are shifted maps, which show geo-
graphic maps inside the nodes of a node-link diagram [OHN*18]
(Figure 15). Nesting in shifted maps allows this technique to pro-
vide for a seamless transition between a geographic placement for
nodes, and a geographic distortion that renders the maps’ visual
(Euclidian) distances between the nodes on the screen depending
on node connectivity using a force-directed layout.
Nesting creates compact representations and shows specific in-
formation about specific regions or nodes. However, there is usu-
ally limited space for the nested representations as nesting requires
space to render the nested views which can then occlude informa-
tion in the background (in the case of geography). Alternatively,
nested views can get quite small. Moreover, extra efforts are re-
quired to understand the context, e.g., the distortion in OD Maps
or the different scales of geography in [BKA*16;OHN*18], and to
provide for a holistic view in the viewer’s mind.
submitted to COMPUTER GRAPHICS Forum (3/2021).

14 S. Schöttler, Y. Yang, H. Pfister & B. Bach / Geospatial Network Visualization
(a) (b)
Figure 15: Example for COMP–Nested: Shifted maps [OHN*18]
nesting geography (mapped) into network nodes (explicit) and pro-
viding seamless transition between (a) mapped geography and (b)
distorted geography showing travel time between locations (images
via shifted-maps.com).
8.4. D3—COMP: Integrated
A technique represents an integrated view of network
topology and geography if it is not possible to separate
network topology and geography into two clearly dis-
tinct visualizations. The visualizations of both geogra-
phy and network topology depend on each other in terms of layout
or other visual mapping decisions. In other words, an integrated
technique cannot be shown in a juxtaposition or superimposition.
For example, the chord diagram in Figure 16(a) [Hen13] can
explicitly show the network topology, while nodes are grouped,
placed, and abstractly color-coded according to their geographic
locations. Geographic information is integrated into an ex-
plicit network visualization. Another example for integration are
Xiao et al.’s Kriskograms [XC09] that show geographic informa-
tion as node labels. The ordering of the nodes is a one-dimensional
projection (not a map projection) of their geographic locations.
Network information can also be represented by modifying a ge-
ographic representation. For example, Alper et al. [ASB07] use a
distorted geographic representation to represent the network infor-
mation (Figure 16(b)).
Integration reduces visual clutter even more than juxtaposition,
as it does not need to split the display space between two compo-
nents. However, in all integrated techniques we found, one of the
components (either GEO or NET) is visually abstracted. Thus, peo-
ple may find it difficult to interpret the abstracted component.
In summary, the COMP dimension describes a spectrum of com-
binations of visual representations for GEO and NET (from loose to
strong integration). Juxtaposition reduces visual clutter by placing
geographic representation and network representation side-by-side.
But extra mental effort is needed to link them visually. Superimpo-
sition overlays one representation on top of another (usually NET
on top of GEO). The two representations are intuitively linked, but
the overlapping of visual elements usually introduces visual clutter.
Nesting embeds one representation inside the visual elements of the
other one, which results in compact designs. Integration fully uti-
lizes the display space by combining the visual representations of
GEO and NET. However, the abstraction introduced requires more
effort to interpret the visual representation.
9. D4: Interactivity (INTERACT)
In addition to the construction of the visual representation through
geography, network and their composition as described in the
(a) (b)
Figure 16: Examples for COMP–Integrated: (a) Nodes are ar-
ranged in a circle for a clear network representation, but grouped
geographically [Hen13], (b) distortions of the globe are used to
visualize the network connectivity [ASB07].
previous sections, interactivity can be essential for visual explo-
ration. This dimension classifies techniques related to understand-
ing geospatial networks, i.e., not including generic techniques such
as pan and zoom, or highlighting visual marks for details. We clas-
sified techniques and papers along a scale from least to most inter-
active, with interaction being either not required,required, or the
technique being interaction only.
We decided to include such pure interaction techniques into this
survey because they provide useful cases about how to address spe-
cific problems in geospatial network visualization; as any interac-
tion technique requires a visualization to function on, the respective
papers (or combined interaction+visualization technique) can still
be classified under all of our other dimensions. Papers that did not
explicitly mention any interaction were classified as interaction be-
ing not required.
9.1. D4—INTERACT: Not Required
Out of the surveyed techniques, 65% do not require
user interaction. In many cases, this means that the
technique generates a static image, which can option-
ally be enhanced with basic interactions like pan and
zoom, scaling (parts of) the visualization up or down, expanding
and collapsing parts of the visualization, or filtering for a subset
of the data. While any technique in this category is usable without
interaction, the extent to which it is useful varies among the tech-
niques and between different data sets. In general, techniques that
do not require interaction have two major advantages: firstly, com-
putational complexity is less of a concern if the image only needs
to be rendered once. Secondly, non-interactive visualizations can
be used in contexts where user interaction is not possible, such as
in print materials or displayed on public screens.
Some spatial generalization techniques [e.g., Guo09;AA11;
CKS*16], i.e., techniques that derive a less detailed visualization
from a more detailed one, can be used to create one-off general-
izations for a display at a specific size and scale, but are also com-
putationally efficient enough to be used as part of an interactively
scalable map or other visualization that automatically updates the
level of generalization based on the zoom level.
Edge bundling techniques also typically do not require interac-
tion, though they can be enhanced with interactive features to ex-
plore bundles. The most common interactive tool for edge bundling
submitted to COMPUTER GRAPHICS Forum (3/2021).

S. Schöttler, Y. Yang, H. Pfister & B. Bach / Geospatial Network Visualization 15
techniques is relaxation, which lets the user interpolate between
the bundled and unbundled views [LHT17b]. In addition to this
generic technique, the creators of the SideKnot edge bundling tech-
nique [PLCP12] explicitly discuss the possibility of integrating
the technique with the EdgeLens interaction technique [WCG03],
which we include under Interaction Only.
Not all non-interactive visualizations are necessarily static im-
ages; animated link textures as proposed by Romat et al. [RAB*18]
use animation to move particles along a network’s links to indicate
link weight (flow). Again, interaction is not required but could be
added for playing and controlling the animations.
9.2. D4—INTERACT: Required
Techniques are labelled as required if the visualization
requires user interaction to fully explore the data set
because some parts of the data are hidden in the de-
fault view, e.g., to reduce visual complexity (24% of
surveyed techniques).
For example, in three-dimensional globe representa-
tions [CEH96;ASB07;LBA10b] (Figure 11(b)), or in the
GeoTime system [KW05], which uses a space-time cube, user
interaction is required to navigate 3D space through rotation
and view the complete visualization. Without interaction, the
visualization hides important information (e.g., data on the other
side of the globe). If the visualization requires configuration, it
requires interaction. Most interactive visualizations are displayed
on a regular screen, limiting interaction modalities to using a
mouse or touch screen. Virtual reality, on the other hand, can offer
rich interaction methods [YDJ*19] (Figure 4(a)).
Interaction is also often required in dealing with large data sets.
Several techniques present the user with an abstract overview and
let them selectively expand parts of the visualization. Li’s module-
based visualization aggregates nodes into meta nodes, which can
be expanded to view the underlying network structure [LBW17b].
Hadlak et al. initially present an aggregated node-link diagram su-
perimposed on a map, then let the user open up additional vi-
sualizations such as matrices to further explore subsets of the
data [HSS11]. The opposite approach is to initially present the user
with a detailed, possibly cluttered, visualization, from which they
can select regions of interest that are then visualized in a more ab-
stract, overview-type visualization [vdEvW14] (Figure 8(a)).
A second group of techniques rely on interaction as an explo-
ration tool, e.g., through filtering the data [VFAA17], or selecting
subsets of the data, which are then displayed in separate, superim-
posed or juxtaposed visualizations [HSS11;LMYH11]. The Flow-
strates technique offers rich interaction capabilities, for example
letting the user interactively select which origin and destination
locations should be displayed in the central heatmap [BBBL11].
Some techniques create a visualization entirely based on user in-
put, for example metro maps that are arranged with a focus on the
user’s travel route [WTLY12].
A third application of interactivity is morphing between differ-
ent network layouts, which can be useful in relating a network
to its geographical context. An example is OD morphing, which
morphs a node-link diagram between an edge-bundled display and
a geographically accurate one, where links are routed along their
true route [LLC*19]. Another example are shifted maps [OHN*18]
(Figure 15), where users can switch between different node-link
layouts based on a mapped representation, and alternative layouts
showing travel distance, travel time, and travel frequency.
9.3. D4—INTERACT: Interaction Only
Interaction Only describes techniques that are not vi-
sualization techniques in their own right, but pure in-
teraction techniques. As such, they are intended to be
applied on top of a visualization created with an ex-
isting technique. The majority of Interaction Only techniques are
designed to be applied to mapped node-link diagrams and mitigate
issues node-link diagrams frequently suffer from, such as overlap,
link clutter, and varying densities.
A number of different lenses have been proposed to help man-
age clutter in superimposed compositions. The Fisheye lens distorts
the entire visualization to allow for zooming into a region while still
viewing the entire network [BMS93]. The EdgeLens [WCG03] and
its 3D counterpart, the 3DArcLens [DSD14a], push links away from
the cursor by bending them. This makes it possible to see nodes and
the underlying map more clearly. An application-specific lens is
provided by the Focus+Context metro map, which combines a de-
tailed view of a small region of the metro map with an aggregated,
simplified view of the larger network context [WC11].
There are techniques that give the user fine-grained control
over moving and displaying links.Link plucking [WC07] lets
the user drag groups of links to the side to reveal what is under-
neath. Riche et al. [RDLC12] present three techniques: interactive
bundling, which lets the user select links to bundle together; link
fanning, which lets users select a node around which links are then
spread out so the user can see individual connections; and link mag-
nets, to be placed by the user, which attract links in their vicinity
(Figure 17(b-d)).
Lastly, there are techniques focused on navigation:Link slid-
ing [MCH*09] snaps the cursor to a link and while dragging the
(a)
(b) (c) (d)
Figure 17: Examples for INTERACT–Interaction Only: (a) Map
insets for navigation: the most relevant nodes are selected and dis-
played as insets [GRE11], (b-d) Illustrations of techniques by Riche
et al. [RDLC12]: Interactive link bundling, fanning, and magnets.
submitted to COMPUTER GRAPHICS Forum (3/2021).

16 S. Schöttler, Y. Yang, H. Pfister & B. Bach / Geospatial Network Visualization
mouse, slides the field-of-view along the link until it reaches the
other end of the link. The same paper introduces another technique:
Bring & Go, which, upon selecting a node in a network, brings all
its direct neighbors into view, even if they are not located in the
current zoomed-in view. The user can then navigate to any of the
connected nodes by selecting it. Ghani et al. [GRE11] present a
navigation technique based on small map insets (Figure 17(a)). A
degree-of-interest function is used to determine which out-of-view
nodes are most relevant; the selected nodes are then displayed in
small map insets along the boundary of the map.
In summary, some visualization techniques propose solutions to
reduce clutter in a static way, e.g., by bundling edges or distort-
ing space to try to address a specific set of tasks. Such techniques
are good for static visualizations (e.g., posters) or contexts where
interaction is tedious or unlikely. However, they make decisions
for the observer with regard to what information to show and in
how much detail. Other techniques introduce interaction to allow
for exploration of e.g., cluttered visualizations, or filter for relevant
information. This aims to support a broader range of tasks than
static visualizations. Eventually, a range of interaction techniques
(Interaction Only, Section 9.3) can potentially be integrated into
visualizations to provide for specific exploration tasks, e.g., Link
Sliding or the EdgeLens can be applied to Kriskograms,OntoTrix,
or flow maps on a 3D globe alike.
10. Addressing Specific Challenges
While Sections 5to 9discussed techniques based on their design,
i.e., how they visualize and abstract information (GEO,NET), how
these representations are composed (COMP), and to what extent a
technique requires user interaction (INTERACT), this section pro-
vides a complementary view on visualization techniques.
Based on earlier versions of our taxonomy (Section 4, Versions
#1 and #3), we discuss common challenges in visualizing spe-
cific types of geospatial networks and to what extent these chal-
lenges are addressed by existing techniques. For each of the tech-
niques included in this survey, we assess whether it is suitable for
thirteen additional data attributes or characteristics:directed
links, weighted links, additional link attributes, additional node at-
tributes, exact point locations, area locations, co-located nodes,
dense networks, networks with varying density, dynamic networks,
uncertain locations, uncertain network topology, and uncertain ad-
ditional attributes. As such, this section does not aim to provide
a comprehensive list of challenges but to provide a practical re-
source for finding the most commonly encountered challenges and
existing approaches to address these challenges when visualizing
specific types of geospatial networks.
10.1. Link and Node Attributes
Geospatial networks can have a variety of attributes associated with
nodes and links. We consider four specific additional data attributes
here: Most common are directed links and weighted links, which
are usually visualized as node-link diagrams making use of arrows
of different thickness (e.g., in flow maps). However, there can be
node and link attributes beyond direction and weight.
Directed links: Out of the additional data attributes, directed
links have received the most attention, most often in the context of
visualizing flows between locations. In node-link diagrams, com-
mon options include arrows, tapered links, and animated links, al-
though there is no clear consensus on which design is preferable,
especially in a geospatial context. Holten et al. conducted a series
of studies [HvW09a;HIvWF11] comparing different encodings for
networks in general and recommend the use of tapered and ani-
mated compressed links. Koylu et al. [KG17] compared tapered
links with arrows. They found that tapered links are only better
for identifying long distance links. Jenny et al. [JSM*18] evaluated
different designs based on tasks including reading flow magnitude
and counting the degree of nodes. They found arrows outperform-
ing tapered links significantly and provide three potential reasons:
1) the gradients of tapered links are inconsistent due to different
link lengths, which is confusing; 2) links in flow maps are usually
thin, resulting a very weak gradient; and 3) incoming flows are dif-
ficult to determine with tapered links.
An alternative to node-link diagrams are matrices. For example,
MapTrix [YDJ*19] juxtaposes a matrix whose rows and columns
are linked to two maps, one for incoming and one for outgoing
flows. A matrix naturally visualizes direction since its rows and
columns typically indicate origins and destinations, respectively.
Weighted links: Another common attribute of links is weight,
i.e., a simple numerical value associated with each link (e.g.,
trade volume). Link weight in node-link diagrams is often vi-
sualized through width of the links. In geospatial networks, this
can lead to obstructing important geographic and other informa-
tion. Using color-gradient encoding (light to dark) compared to
link width has been found more effective and efficient in a con-
trolled user study [DWCM18] for small geospatial networks; par-
ticipants commented that maps using color gradients are clearer
than maps using line thicknesses. As an alternative to static maps,
Romat et al. [RAB*18] use animated link textures to indicate link
weight, while keeping all links to the same width. Their user study
found that participants could discriminate up to six different values
based on animated link textures, but that the use of particle pat-
terns resulted a poor performance for a quick estimate yet was the
best choice for accessing details. However, no comprehensive and
comparative study exists so far.
Additional link attributes: In cases where links have additional
attributes beyond simple weights and directions, colors are com-
monly used to encode categorical information. For example, Abel
and Sander [AS14] use color in links to represent regions of the
countries in global migration. In transit maps, colors are usually
used to represent different transit lines [WNT*20].
Additional node attributes: In many cases, nodes are repre-
sented as identical shapes (e.g., circles). However, there are sce-
narios that require representing additional attributes on the nodes.
One common case is using the size of the node to encode a quan-
titative value. The quantitative value can be the net in/out flow into
the geographic location the node represents, or a property asso-
ciated with the geographic location but irrelevant to the network
topology (e.g., the population of a country [SV10]). When it comes
to the need of encoding more complex data on nodes, glyphs have
been commonly used. For example, Yang et al. [YDGM17] place
two semicircles on one location: one half represents net inflow and
the other half represents the net outflow. Statistical charts can also
be nested into nodes to visualize a variety of additional informa-
submitted to COMPUTER GRAPHICS Forum (3/2021).

S. Schöttler, Y. Yang, H. Pfister & B. Bach / Geospatial Network Visualization 17
Figure 18: Node-overlap removal in the ‘Vistorian’: Overlapping
nodes are expanded circularly, using a simple range slider to set
the expansion radius of the node circles. Links between nodes in
the same position are shown inside the circle [BRF*15].
tion [vdEvW14]. If node size is used to encode quantitative values
on a mapped geography representation, small and close regions will
pose a problem due to glyphs overlapping other glyphs, links, and
potentially information on the underlying map.
10.2. Geographic Locations
Our definition of a geospatial network includes networks associ-
ated with a variety of types of geolocations. We consider three dif-
ferent data type-related challenges. A rough division can be made
into point locations (0D) and area locations (2D). For both of
these location types, a multitude of techniques exist, with many
techniques usable for both types. However, the distinction mat-
ters when addressing a common challenge related to locations: co-
located nodes, i.e., multiple nodes at the same location (e.g., com-
panies in the same building).
Co-located nodes are a challenge because placing multiple nodes
in the same position leads to occlusion not only of the nodes, but
also of potential links between them and surrounding geographical
information. For near-identical positions, some distorted geogra-
phy representations (see Section 5.2) can be used to mitigate this
issue to an extent, though distortion cannot address exactly identi-
cal positions. The Vistorian [BRF*15] orders nodes from the same
location on a circle around their position while the user interac-
tively controls the circle radius (Figure 18).
When multiple nodes are located in the same area (e.g., a coun-
try), they could be juxtaposed inside this area, although we could
not find a technique implementing this. Nodes might even be placed
in an intelligent way that minimizes link crossings.
An approach that can theoretically be used for both point and
area locations that are co-located is to detach the network topol-
ogy from the precise geography to an extent. This can be achieved
by using abstract or distorted geography representations, e.g.,
showing locations by grouping nodes occupying the same space
into segments along a circle [AS14;Hen13], a line [XC09] or
a matrix [BPLL11] (abstract) or compositions such as Necklace
maps [SV10] (distorted), which arrange nodes on a circle super-
imposed on a map, with node positions based on geolocations as
shown on the map. Rather than abstracting the geography, an alter-
native approach can be the use of a juxtaposed composition, which
produces a looser integration of geography and network. An exam-
ple is the MapTrix technique [YDGM17], which uses a matrix to
display the network. Leader lines connecting the matrix to a juxta-
posed map are used to link geography and network; these can easily
link multiple matrix rows or columns to the same location.
The inverse challenge to co-located nodes would be nodes with
multiple positions (e.g., if an organization has multiple offices in
different locations). However, we are not aware of any techniques
addressing this specifically for geospatial networks, or discussion
of this challenge elsewhere. In non-geographic networks, this has
been addressed by node-cloning [AvHK06], but this technique has
not been adapted for geospatial networks and comes with its own
set of challenges (e.g., showing which nodes are the same, deciding
how to draw or duplicate links).
An additional potential challenge we see are node locations at
different granularities, i.e., a network where some nodes are asso-
ciated with geolocations at the level of countries, others with cities
or precise street addresses. Similarly to the previous challenge, this
is not a commonly discussed challenge, and we are not aware of
any techniques addressing this.
10.3. Link Density
In the context of link density, we consider two main data character-
istics: Firstly, dense networks have too many links that cause vi-
sual clutter when visualized in node-link and arc diagrams. Super-
imposing a node-link diagram on a mapped or distorted geography
representation will clutter the display and can make geography as
well as the network unreadable. A different, but related challenge is
posed by networks with varying density, where some geographic
locations feature significantly more nodes and links than others.
A common solution for the visualization of dense non-geospatial
networks are matrices, which scale to complete networks, i.e.,
networks where each node is connected to all other nodes
(density=1.0). Matrices have been extensively studied in visual-
ization, starting with Ghoniem et al. [GFC04] who found that
matrices scale well to larger network data. Matrices have subse-
quently been found more efficient for weighted and undirected net-
works [ABH*13]. In a crowdsourcing study, Okoe et al. [OJK18;
OJK19] found that node-link diagrams worked better for topol-
ogy and memorability tasks, while matrices were better for cluster-
related tasks. However, it is challenging to integrate geographic
information into a matrix. One solution is a juxtaposed com-
position in which leader lines connect rows and columns of a
matrix to their associated geographic locations, shown on two
maps [YDGM17] or grouped by location [BPLL11]. In a user study
by Yang et al. [YDGM17], these two designs had similar perfor-
mance and scaled better than bundled node-link diagrams superim-
posed on maps. An alternative solution is the nested composition
of Wood et al.’s OD Maps [WDS10], in which the entire map is du-
plicated and nested into itself to resemble a map and thus represent
the geographic context.
For applications where superimposed node-link diagrams are
required, different methods exist to make these visualizations
more readable and navigable even at high link densities. Edge
bundling [e.g., HvW09b;LHT17a] or edge routing [BS15] can
free visual space by visually grouping links, though this results
in a loss of accuracy when following individual links [BRH*17].
Less problematic are techniques that keep links separate, such as
manually routing links through ‘magnets’ [RDLC12] or automat-
ically routing links around each other [BST00;BW98;JSM*17].
Two comprehensive studies found that curved links performed
better than straight ones when superimposed on maps [JSM*18;
submitted to COMPUTER GRAPHICS Forum (3/2021).

18 S. Schöttler, Y. Yang, H. Pfister & B. Bach / Geospatial Network Visualization
DWCM18]. One of them, Jenny et al. [JSM*18] conclude that well-
designed curved links can i) minimize the overlaps between links,
ii) avoid them intersecting at acute angles, and that iii) curved links
should be curved as little as possible.
Aggregating nodes is another option as it automatically leads
to link aggregation as well. Examples are Li et al.’s module-based
visualization [LBW17b], in which dense clusters of the network
are visually represented by aggregate symbols, as well as the use of
hierarchical clustering to aggregate nodes by Zhu and Guo [ZG14].
Moreover, rather than aggregating nodes or links individually,
the entire network can be abstracted such that higher-level fea-
tures of the topology are shown. Andrienko et al. [AA11] intro-
duce an algorithm to generalize and aggregate directed geospa-
tial networks such that a simpler version of the network is pro-
duced, which can then be visualized. A slightly different approach
is Guo and Zhu’s flow data smoothing [GZ14], which displays
link-like arrows, but they are not connected to fixed nodes. Pat-
tern maps [YWZ*19] indicate directions of flows in space through
a hexagonal pattern; another technique by Kim et al. [KJW*18]
uses a field line-like visualization for the same purpose.
While most of the above-mentioned techniques for reducing
link clutter work globally on the entire network, other techniques
have a local scope in that they resolve clutter for some nodes and
links: dynamic edge lenses [Zou16], temporarily pushing away
links [DSD14a], or topological fisheye techniques [BMS93]. Addi-
tionally, there are many generic graph interaction techniques which
could be used for geospatial networks, such as the Local Edge
Lens and Bring Neighbors Lens [TAvHS06]. These local interac-
tion only techniques are good for keeping the overall visualization
stable while exploring subgraphs and specific geographic regions.
Also, while globally applied techniques can help to deal with non-
uniform densities, local approaches like these do not unnecessarily
sacrifice detail in less dense areas.
Finally, there are several approaches specifically useful for dense
areas in networks with non-uniform densities, i.e., networks that
also contain much sparser areas. One solution is using distortion
to increase the size of dense areas, for example, through centrality-
based scaling [MG06]. Metro map layouts [HMdN06] address this
issue as well and could potentially be applied to non-transport
networks as well. Interaction techniques that let the user view
separate visualizations for subgraphs are also a possible solution
[HSS11]. The problem can be avoided using an abstract geographic
representation [AS14] or matrices to avoid positioning nodes based
on their geographic location. Another solution to varying densities
can be map insets [GRE11;BKA*16] which scale up particularly
dense regions to reveal the network topology and are additionally
able to bring subgraphs geographically closer, distorting geogra-
phy.
10.4. Dynamic Geospatial Networks
Studying dynamic networks, i.e., networks that change over time
as nodes and links appear or disappear or link weight changes
over time, has led to many techniques, nicely summarized by
Beck et al. [BBDW14]. Geospatial networks can involve a range
of additional types of changes such as nodes changing positions
and geography changing structure (e.g., when countries merge or
split). While visualizations for geo-temporal (non-network) data
have been summarized by Bach et al. [BDA*17], little is known
about visualizing dynamic geospatial networks.
One common approach is to use animation to show changes.
Juxtaposition of small multiples, one for every time step,
is another straightforward method explored and evaluated by
Boyandin et al. [BBL12]. Their study, comparing animation with
small multiples, found that animation led participants to more find-
ings with local events and changes between subsequent years.
However, small multiples led to more findings concerning longer
time periods. Animation can also be used with other types of visual
representations, for example animating deformations of a 3D globe
to show how strongly linked different locations are [ASB07].
For smaller networks with a moderate amount of time steps,
where only the weights are dynamic, the Flowstrates tech-
nique [BBBL11] can be used. This technique juxtaposes two maps
and a custom time series visualization such that the time series of
each link is shown in the center, connected to start and end location
on the two maps respectively. For some networks, using the third
dimension to represent time, i.e., displaying the network in a space-
time cube, may also be a solution [KW05]. For exploratory visual-
ization, Hadlak et al. [HSS11] propose the concept of in-situ explo-
ration of dynamic networks, where users can display and modify a
variety of additional visualizations.
In cases where it is not necessary to explicitly show changes
over time in sequence, temporal abstraction or summarization can
be an alternative approach. Andrienko et al. [AAFW17] propose
a technique to abstract dynamic network data over space and
time, showing the result in a glyph-based visualization. Shifted
maps [OHN*18] is a technique intended for personal movement
data which includes network layouts based on how frequently
someone travels to certain locations or how long the journey takes.
10.5. Uncertainty
Visualizing uncertainty in networks is an underexplored, but grow-
ing, field of research. In the context of geospatial networks, uncer-
tainty can occur in different ways: Firstly, there is the issue of un-
certain network topology, i.e., uncertainty about whether a certain
node or link exists or what direction a directed link has. Secondly,
there can be uncertain locations, i.e., where exactly a node is lo-
cated, or in the case of geolocated links, what exactly their trajec-
tories are. Lastly, nodes and links can have uncertain additional
attributes. Uncertainty can be seen as ranging on a spectrum from
exactly known via uncertain to unknown, and the grade of uncer-
tainty will affect how it can best be addressed in a visualization.
We did not find any concrete techniques specifically for visual-
izing geospatial networks with uncertain topology or geolocations.
However, Von Landesberger et al. [LBW17a] have proposed a ty-
pology for uncertainty in geospatial graphs, and identified many ap-
proaches to visualizing uncertainty that have not been explored or
tested yet. They focus on modifying geolocated node-link diagrams
to visualize uncertain graphs and thus do not consider solutions
outside of what we call explicit geography and network representa-
tions. Also, techniques with an abstract geography representation,
e.g., by using spatially ordered chord diagrams [Hen13;AS14],
Kriskograms [XC09], or matrices [BPLL11], can be a solution for
submitted to COMPUTER GRAPHICS Forum (3/2021).

S. Schöttler, Y. Yang, H. Pfister & B. Bach / Geospatial Network Visualization 19
uncertain locations, since these techniques do not rely on precise
geospatial locations to position elements of the visualization. The
probabilistic graph layout introduced by Schulz et al. [SNG*17]
is a technique to visualize networks with uncertain link weights
or other numerical link attributes. The technique creates a ‘blurred’
node-link diagram by decomposing the uncertain graph into its pos-
sible instances, then visually recombining them using node splat-
ting and edge bundling.
Overall, there is a clear need for a larger variety of techniques for
more different use cases here. In particular, there is a lack of tech-
niques usable for data exploration. Uncertainty in dynamic geospa-
tial networks is another area that has not been addressed.
11. Discussion
This survey identified 95 papers presenting techniques for geospa-
tial network visualization, to which a structured coding methodol-
ogy was applied. In this section, we discuss our design space, how
to make trade-offs between techniques along our dimensions, and
which issues remain future work.
11.1. Design Space
Our design space (Sections 5–9) provides an overview of various
techniques’ designs with the goal of providing a conceptual un-
derstanding for possible design solutions. Following several iter-
ative steps to classify techniques according to different schemata
(Section 4.2), we eventually defined five design dimensions GEO,
NET:NODE,NET:LINK,COMP, and INTERACT. This design space
was created on the premise that each visualization technique has
to solve a trade-off between showing different levels of detail and
precision for different types of information in order to deal with the
complexity of data in geospatial networks.
To that end, each of the dimensions GEO,NODE, and LINK
describes a spectrum to categorize techniques according to how
much information they show explicitly and how much information
about the data gets abstracted and aggregated. We classified tech-
niques from each paper into rough categories along these dimen-
sions. This allows us to capture essential steps along the dimen-
sions and their individual characteristics. Our dimensions COMP
and INTERACT exhibit more discrete categories along their re-
spective dimensions—loose to tight integration for COMP and not
required to interaction only for INTERACT since these two dimen-
sions naturally fall into more discrete steps.
We want to emphasize that all our dimensions are to be under-
stood as a continuum between two respective poles. Such a contin-
uum has three benefits:
•Our design space is high-level and expresses ideas and trade-
offs, rather than just specific existing solutions. This implies
that there is a huge set of design solutions and techniques to ex-
plore the trade-off between explicit and abstract representations
and we believe such a spectrum is a powerful thinking tool when
discussing existing and designing new techniques. For example,
when searching for a given technique to apply to a specific appli-
cation problem, the designer or analyst can use each dimension
to assess the relative importance and of that information (net-
work topology with nodes and links, or geography). Then, they
could look for specific techniques or inform their own design
trying to include as much information from each dimension as
they need.
•Thinking of our dimensions as continuous makes them less
rigid and open to capturing new techniques in the future.
A new technique can be placed onto these dimensions, poten-
tially creating a new category, or requiring the refinement or
splitting of an existing category. Our categories represent groups
of techniques with common characteristics along these dimen-
sions. They help capturing these high-level characteristics (e.g.,
distortion) and to orient the user of our survey and design space.
•Continuous dimensions allow techniques to float along the
continuum and exploit a richer set of designs. For example,
edge bundling captures many different individual techniques,
each of which potentially floats along the LINK-dimension. For
example, one edge bundling technique can show explicit links
if the bundling parameter is low, but aggregate individual links
and make them harder to trace under another parameter setting.
Consequently, a single technique (edge bundling) can occupy an
entire segment in our dimension.
In that sense, our design space is complete for the papers we
could find, and we believe it is open enough to capture a range of
future techniques.
11.2. Design Space Coverage and Open Designs
Our design space allows us to discuss existing and common designs
(Figure 19), less common and underexplored designs, drawbacks of
individual designs, trade-offs between designs along the same di-
mension, as well as to negotiate trade-offs between design solutions
across dimensions.
Common combinations in our design space include Mapped
×Explicit Nodes ×Superimposed (53%) as well as Aggregated
Links ×Superimposed (27%). The latter contains nearly all edge
bundling techniques as well as flow map layout techniques. In con-
trast, other combinations have only a few associated techniques,
such as Abstract Nodes ×Abstract Links (2%). We believe that
many of the rarer combinations are underexplored but possible; for
example, we can imagine abstract nodes juxtaposed with a mapped
or distorted geography representation. We also believe there is a
large unexplored design space of juxtaposed and linked visual-
izations where the network can be represented in a clearer way
while the geography is mapped.MapTrix [YDGM17] and Flow-
strates [BBBL11] are examples of techniques in this space.
However, our design space is descriptive, rather than prescrip-
tive. In other words, we can point to holes in the design space, e.g.,
we could not find any techniques for a combination of juxtaposi-
tion and abstract geography. However, we cannot prescribe how a
new technique should be designed. This is different from prescrip-
tive design spaces which can be used to create new techniques by
combining a set of options (e.g., layout, interaction technique, color
encoding, etc.). The dimension with the most prescriptive power is
COMP as our categories (juxtaposed,superimposed,nested, and in-
tegrated) are quite distinct and generic design solutions. However,
the categories for nesting and integration, for example, both capture
a range of possible techniques and are hence less useful to describe
a specific visual design. Likewise, while juxtaposition and super-
imposition present very specific design solutions, their individual
submitted to COMPUTER GRAPHICS Forum (3/2021).

20 S. Schöttler, Y. Yang, H. Pfister & B. Bach / Geospatial Network Visualization
Figure 19: Overview of techniques classified in our design space. This visualization illustrates the trade-off effect between GEO,NODE, and
LINK: there is little correlation between the abstract ends of these dimensions, i.e., most techniques abstract only one dimension.
implementation offers a wide range of design parameters. For ex-
ample, a juxtaposition needs to clarify the size of each view (e.g.,
large geographic view and small network topology if the network is
rather small but spatially unequally distributed), as well as mecha-
nisms for connecting information across both views (e.g., brushing
and linking, visual links, or consistent visual encodings).
11.3. Discussing Techniques
Our design space makes it straightforward to discuss conceptual
advantages and drawbacks of specific designs. Techniques on the
explicit end of the dimensions show more explicit and, as a conse-
quence, often more detailed information. Generally, techniques on
the abstract end show less detail and aggregate and abstract infor-
mation appropriate to specific tasks. Less information makes tasks
related to individual nodes, links, and geographic locations harder
or impossible. Generally, abstracting information is useful in two
scenarios: i) support tasks that require abstractions (e.g., analyze
nodes and connectivity per geographic region) and ii) hide infor-
mation less important to a task and that would otherwise clutter the
interface.
GEO:Mapped techniques such as a 3D globe or 2D projected
maps offer high precision regarding geography but pose constraints
on the position of nodes—if used in a superimposed or integrated
way—or require linking information between views if presented in
a side-by-side view. 3D globes also require navigation. In general,
distortion is an inevitable consequence of transforming a sphere or
an ellipsoid onto a plane. It is impossible to unfold the Earth onto a
planar map without distortion [Sny87;JŠA*17]. This can result in
exaggerated and unequal link lengths for superimposed node-link
diagrams, but also can result in a false perception of where a link is
located, if that matters. Using three-dimensional globes [CEH96;
KB16;YJD*18;YDJ*19] partly addresses this issue, but virtual
reality is not always accessible.
NODE and LINK:Explicit network topology (nodes and links)
combined with explicit geography is required in tasks that re-
quire parallel understanding of network topology (node connectiv-
ity, path following, clusters, identifying network motifs [LPP*06])
and precise geography at the same time. It can also precisely pre-
serve the geography. However, the resulting views are usually visu-
ally cluttered.
COMP:A loose integration (juxtaposition) is flexible, if not ag-
nostic, to the respective representations of geography and network
topology. Both facets can employ a visualization technique that best
supports the information shown and the task at hand. As mentioned
earlier, juxtaposition requires mechanisms to relate information
from both views. Superimposition represents a stronger integration
as network and geographical information are displayed in the same
visual space. Most commonly, nodes are placed at their associated
geographic positions. This might support tasks related to nodes
and their positions as well as questions of local connectivity. On
the downside, superimposition comes with the common issues that
motivated this survey: node overlap, long links, ambiguous links,
visual clutter, and the general difficulties to understand network
topology. A visual solution includes moving nodes away from the
map, presenting a transition to a juxtaposed view [SV10]. Interac-
tive techniques (e.g., EdgeLens [WCG03], Bring & Go [MCH*09]
or node circles [BRF*15]) can further alleviate these issues. Some
network representations, e.g., adjacency matrices, are hard to su-
perimpose since node positions are constrained to vertical and
horizontal rows and columns. Hybrid techniques such as Node-
Trix [HFM07] superimposed on a map could provide a solution
for analyzing local clusters and inter-regional connections. Nested
and integrated techniques provide for a tighter connection of topo-
logical and geographic information, but the tight integration also
reduces flexibility in choosing geography and network representa-
tion separately.
INTERACT:Without interaction, a visualization needs to be
very careful about clutter reduction and can present only a single vi-
sual representation. Interaction could allow for ‘moving along’ the
dimensions in our design space, e.g., by morphing from a mapped
via a distorted to an abstract geographic representation. A range of
interaction techniques have been created to address specific issues.
Interaction for exploring geospatial networks is a largely underex-
plored space in which we see great potential for novel techniques.
11.4. Negotiating Trade-Offs
Knowing our design space and the individual drawbacks of each
category along the dimensions helps negotiate trade-offs between
each category as well as think about hybrid techniques. For exam-
ple, one possible way to address the trade-off between geography
and network topology is to morph between geographic and force-
submitted to COMPUTER GRAPHICS Forum (3/2021).

S. Schöttler, Y. Yang, H. Pfister & B. Bach / Geospatial Network Visualization 21
directed layouts, or to restrict geographic information to the node
neighborhood [OHN*18]. Another possible trade-off is geographic
distortion [BDD*16;ASB07] whereby through sophisticated ge-
ometric transformations geographic locations are relaxed and dis-
torted to bring nodes closer to their positions in a force-directed
layout. Hybrid approaches in terms of positioning, such as the ap-
proach taken in metro maps, are another way of giving more im-
portance to the network topology.
11.5. Limitations and Possible Extensions
Some techniques in our collection caused more discussion when
classifying than others. While mapped techniques with explicit
nodes and links are easy to classify—perhaps due to the tradition
our community has with node-link diagrams—techniques aggre-
gating nodes and links were less common and consequently caused
us more discussion in understanding and capturing their idea. Other
ambiguous cases are found between distorted and abstract geogra-
phy, as some abstract designs, like spatially ordered nodes on a
circular chord diagram, could also be seen as extreme distortions.
Still, we see the flexibility of our design space as a strength.
Future analyses of our collection of techniques may classify
them differently. For example, in Version #2 of our taxonomy (Sec-
tion 4.2), we tried to classify techniques by visualization types for
networks (e.g., matrix, arc diagram). While we found that such a
categorization does not capture the information necessary for our
discussion, we believe such a categorization might still be useful to
inform novel techniques in a more prescriptive way. However, care
must be taken to not restrict creativity and to capture all possible
groupings and combinations of geography and network topology.
11.6. Addressing Open Challenges
Section 10 provided an overview of how specific challenges,
grouped by data type, are addressed in current techniques. Despite
our survey including techniques from 95 papers, there is a large
array of unsolved challenges in visualizing geospatial networks.
For example, among the challenges most under-addressed is uncer-
tainty. While uncertainty in networks is generally under-addressed,
we could not find solutions for visualizing multiple positions, miss-
ing node positions, missing locations on links (trajectories) and the
range of uncertainties present in geographic visualization.
We also found few techniques directly addressing problems in
node overlap and resolving ambiguity when links overlap. Routing
seems to be promising here but future routing algorithms could try
to take geography more into account, to route links, e.g., through
empty space or along semantic trajectories such as rivers or fron-
tiers. Finally, techniques for dynamic geospatial networks, includ-
ing nodes moving between locations, a changing network topology
as well as other changing node attributes are under-explored. Ap-
proaches from spatio-temporal visualization [BDA*17] could help
finding appropriate solutions.
11.7. Towards a Task Taxonomy
Much of our discussion in this survey is making reference to tasks.
Tasks are a powerful concept in visualization that inform design
and evaluation and have been formulated for networks in gen-
eral [LPP*06] and dynamic networks [BPF14;KK17] as well as
for geographic visualization [Rot13]. The latter taxonomy included
objectives (identify, compare, rank, associate, delineate, procure,
predict, & prescribe) and seventeen operators (import, export, save,
edit, annotate, re-express, arrange, sequence, re-symbolize, over-
lay, pan, zoom, re-project, search, filter, retrieve, & calculate).
However, such tasks are similar to generic taxonomies in visual-
ization [AS04] and do not provide the expressiveness necessary
for geospatial networks.
More specifically, Andrienko et al. [AA06] formalized tasks for
spatio-temporal visualizations, considering space, time and objects
as fundamental elements and tasks as queries for the information
associated with them. Yang and Goodwin [YG19] interviewed do-
main experts who analyze geospatial networks in their professional
work and identified three analytical targets: single flow (i.e., a flow
between two geographic locations), total flow (i.e., all flows linked
to a given location) and regional flows (i.e., flows between loca-
tions within a geographic area). However, these works only focus
on single perspectives of our design space.
To the best of our knowledge, a structured task taxonomy for
geospatial networks does not exist yet. Such a task taxonomy could
help to standardize evaluation and allow for comparison across
technique papers, and additionally to characterize systems in terms
of task support, thus allowing for more informed technique selec-
tions for applications.
11.8. Empirical Evidence
Empirical evidence on how specific geospatial network visualiza-
tion techniques perform is sparse (7% in our collection). Existing
papers include mostly case studies, sometimes small quantitative
user studies. The tested tasks are predominantly rather basic, with
more complex tasks rarely being evaluated. Even when there is
a stronger focus on evaluation, comparison across publications is
practically impossible due to the lack of standardization. Our de-
sign space might provide some scaffolding here. For example, open
questions that should be addressed by empirical evaluations include
the effectiveness of geographic distortion and the impact and read-
ability of abstract geographic representations.
12. Conclusion
This survey presents a structured collection of 95 visualization
and interaction techniques for geospatial networks. We explored
various ways of categorizing our techniques and ended up with
five dimensions. Each technique can be described along each of
these dimensions and compared to other techniques. At the same
time, these dimensions provide for a design space to inspire future
techniques. We also discussed common challenges in visualizing
geospatial networks, and whether our collected techniques are suit-
able for addressing these challenges. We concluded with a discus-
sion of the design space, techniques, and a list of directions for
future research and hope this survey will be an inspiration for vi-
sualization designers to find novel and creative techniques to solve
the many open questions as well as a useful guide for analysts,
scientists, and students (yet) outside the field of visualization and
geography.
submitted to COMPUTER GRAPHICS Forum (3/2021).

22 S. Schöttler, Y. Yang, H. Pfister & B. Bach / Geospatial Network Visualization
Acknowledgements
We would like to thank the anonymous reviewers for their valu-
able comments. Yalong Yang is supported by a Harvard Physical
Sciences and Engineering Accelerator Award.
References
[AA06] ANDRIENKO, NATALIA and ANDR IEN KO, GE NNADY.Ex-
ploratory Analysis of Spatial and Temporal Data: A Systematic Ap-
proach. Springer Science & Business Media, 2006. DOI:10.1007/3-
540-31190- 4 3,21.
[AA11] ANDRIENKO, N. and AN DRI ENKO , G. “Spatial Generalization
and Aggregation of Massive Movement Data”. IEEE Transactions on
Visualization and Computer Graphics 17.2 (Feb. 2011), 205–219. DO I:
10.1109/TVCG.2010.44 12,15,18.
[AAB*13] ANDRIENKO, GENNADY, ANDR IEN KO, NATALI A, BAK, PE-
TER, et al. Visual Analytics of Movement. Springer, Berlin, Heidelberg,
2013. DOI:10.1007/978-3-642-37583-5 3.
[AAFG18] ANDRIENKO, GENNADY, ANDR IEN KO, NATALI A, FUCHS,
GEORG, and GARCIA, JOSE MANUEL CO RDE RO. “Clustering Trajec-
tories by Relevant Parts for Air Traffic Analysis”. IEEE Transactions
on Visualization and Computer Graphics 24.1 (Jan. 2018), 34–44. DOI:
10.1109/TVCG.2017.2744322 4.
[AAFW17] ANDRIENKO, GENNADY, ANDR IEN KO, NATALI A, FUCHS,
GEORG, and WO OD, JO. “Revealing Patterns and Trends of Mass Mo-
bility Through Spatial and Temporal Abstraction of Origin-Destination
Movement Data”. IEEE Transactions on Visualization and Computer
Graphics 23.9 (Sept. 2017), 2120–2136. DOI:10 . 1109 / TVCG .
2016.2616404 7,12,13,18.
[AAK*14] ARCHAMBAULT, DANIEL, ABEL LO, JAMES, KENN EDY,
JESSIE, et al. “Temporal Multivariate Networks”. Multivariate Network
Visualization. Ed. by KERREN, ANDR EAS, PURCHASE, HEL EN C., and
WARD, MATTHE W O. Vol. 8380. Series Title: Lecture Notes in Com-
puter Science. Cham: Springer International Publishing, 2014, 151–174.
DOI:10.1007/978-3-319-06793-3_8 2.
[ABH*13] ALPER, BASAK, BACH, BENJAMIN, HENRY RI CHE,
NATHALIE, et al. “Weighted Graph Comparison Techniques for Brain
Connectivity Analysis”. Proceedings of the SIGCHI Conference on
Human Factors in Computing Systems. CHI ’13. New York, NY, USA:
ACM, 2013, 483–492. DOI:10.1145/2470654.2470724 17.
[ACM19] ACM SIGCHI. ACM SIGCHI. 2019. URL:https : / /
sigchi . org / conferences / conference - history / chi/
(visited on 07/19/2019) 4.
[AS04] AMAR, R. and STASKO, J. “A Knowledge Task-Based Framework
for Design and Evaluation of Information Visualizations”. IEEE Sym-
posium on Information Visualization. IEEE Symposium on Information
Visualization. Oct. 2004, 143–150. DOI:10.1109/ INFVIS.2004 .
10 21.
[AS14] ABEL, GU Y J. and SANDER, NIKOL A. “Quantifying Global In-
ternational Migration Flows”. Science 343.6178 (Mar. 28, 2014), 1520–
1522. DOI:10.1126/science.1248676 2,8,9,17–19.
[ASB07] ALPER, BASAK, SÜMENGEN, SELÇUK, and BALCISOY, SE-
LIM. “Dynamic visualization of geographic networks using surface de-
formations with constraints”. Proc. of the Computer Graphics Interna-
tional Conference (CGI), Computer Graphics Society, Petrópolis, Brazil.
2007 7–9,12,14,15,18,21.
[AvHK06] ABEL LO, JAMES, van HAM, FRANK, and KRISHNAN,
NEERAJ. “ASK-GraphView: A Large Scale Graph Visualization Sys-
tem”. IEEE Transactions on Visualization and Computer Graphics 12.5
(Sept. 2006), 669–676. DO I:10.1109/TVCG.2006.120 17.
[Bat09] BATTE RSB Y, SARAH E. “The Effect of Global-Scale Map-
Projection Knowledge on Perceived Land Area”. Cartographica: The
International Journal for Geographic Information and Geovisualization
(Apr. 14, 2009). DOI:10.3138/carto.44.1.33 7.
[BBBL11] BOYAN DIN, I LYA, BERTI NI, ENRICO, BAK, PE TER , and
LALANNE, DEN IS. “Flowstrates: An Approach for Visual Exploration
of Temporal Origin-Destination Data”. Computer Graphics Forum 30.3
(2011), 971–980. DO I:10. 1111/ j. 1467- 8659 .2011 .01946 .
x11,13,15,18,20.
[BBDW14] BEC K, FABIAN, BURCH, MICHAEL, DIEHL, STE PHA N, and
WEISKOP F, DANIEL. “The State of the Art in Visualizing Dynamic
Graphs”. EuroVis - STARs. Ed. by BOR GO, R., MACIEJEWSKI, R.,
and VIOLA, I. The Eurographics Association, 2014. D OI:10 .2312 /
eurovisstar.20141174 2,18.
[BBDZ08] BÖT TGE R, JOAC HIM , BRA NDE S, ULRIK, DEUS SEN, OLIVER,
and ZIEZOLD, HENDRIK. “Map Warping for the Annotation of Metro
Maps”. IEEE Computer Graphics and Applications 28.5 (Sept. 2008).
Conference Name: IEEE Computer Graphics and Applications, 56–65.
DOI:10.1109/MCG.2008.99 8.
[BBL12] BOYAN DIN, I LYA, BERTI NI, ENRICO, and LAL ANN E, DENIS.
“A Qualitative Study on the Exploration of Temporal Changes in Flow
Maps with Animation and Small-Multiples”. Computer Graphics Forum
31.3 (2012), 1005–1014. DO I:10 .1111 /j . 1467- 8659 . 2012.
03093.x 11,18.
[BDA*17] BACH, B., DRAGICEVIC, P., ARCHAMBAULT, D., et al. “A
Descriptive Framework for Temporal Data Visualizations Based on Gen-
eralized Space-Time Cubes: Generalized Space-Time Cube”. Computer
Graphics Forum 36.6 (Sept. 2017), 36–61. DO I:10 . 1111 / cgf .
12804 3,5,18,21.
[BDD*16] BOUTS, Q. W., DWY ER, T., DYK ES, J ., et al. “Visual Encod-
ing of Dissimilarity Data via Topology-Preserving Map Deformation”.
IEEE Transactions on Visualization and Computer Graphics 22.9 (Sept.
2016), 2200–2213. DO I:10. 1109/ TVCG. 2015. 2500225 2,7–9,
12,21.
[Ber83] BERTIN, JAC QUE S.Semiology of Graphics. Madison, Wis: Uni-
versity of Wisconsin Press, 1983. 415 pp. ISB N: 978-0-299-09060-9 3.
[BEW95] BECKER, R.A., EICK, S.G., and WI LKS, A.R. “Visualiz-
ing network data”. IEEE Transactions on Visualization and Computer
Graphics 1.1 (Mar. 1995), 16–28. DOI:10.1109/ 2945.468391 2,
10.
[BKA*16] BRO DKORB, FE LIX, KUIJPER, ARJA N, ANDRI ENKO , GEN -
NADY, et al. “Overview with details for exploring geo-located graphs
on maps”. Information Visualization 15.3 (July 1, 2016), 214–237. DOI:
10.1177/1473871615597077 8,14,18.
[BMS93] BROW N, MARC H., MEEHAN, JAME S R., and SARKAR,
MANOJIT. “Browsing Graphs Using a Fisheye View (Abstract)”. Pro-
ceedings of the INTERACT ’93 and CHI ’93 Conference on Human
Factors in Computing Systems. CHI ’93. New York, NY, USA: ACM,
1993, 516–. DO I:10.1145/169059.169474 8,16,18.
[Bou17] BOUTS, QUIR IJN. “Geographic Graph Construction and Visual-
ization”. PhD Thesis. Eindhoven University of Technology, 2017 7.
[Boy20] BOYANDIN, ILYA.Relocations between Swiss
cantons in 2016 - Flowmap.blue. Accessed: 2021-02-
01. 2020. UR L:https : / / flowmap . blue /
15kwLB4baXZ7jpip8q0JjgR6zDoS5Gt3gMLCTUAboQxk 11.
[BPF14] BACH, BENJAMIN, PIETRIGA, EMM ANU EL, and FEKE TE,
JEAN-DANIEL. “GraphDiaries: Animated Transitions and Temporal
Navigation for Dynamic Networks”. IEEE Transactions on Visualization
and Computer Graphics 20.5 (May 2014), 740–754. DOI:10. 1109/
TVCG.2013.254 21.
[BPLL11] BACH, BEN JAMI N, PIETRIGA, EMMANUEL, LICCARDI,
ILARIA, and LEGOSTAEV, GENNA DY. “OntoTrix: a hybrid visualization
for populated ontologies”. Proceedings of the 20th international confer-
ence companion on World wide web. 2011, 177–180. DOI:10.1145/
1963192.1963283 2,8,9,17–19.
[BRF*15] BACH, BENJAMIN, RICH E, NATHALIE HENRY, FERNANDEZ,
ROLAND, et al. “NetworkCube: bringing dynamic network visualiza-
tions to domain scientists”. Posters of the Conference on Information
Visualization (InfoVis). 2015 17,20.
submitted to COMPUTER GRAPHICS Forum (3/2021).

S. Schöttler, Y. Yang, H. Pfister & B. Bach / Geospatial Network Visualization 23
[BRH*17] BACH, BENJAMIN, RICH E, NATHALIE HENRY, HURTE R,
CHRISTOPHE, et al. “Towards Unambiguous Edge Bundling: Investi-
gating Confluent Drawings for Network Visualization”. IEEE Transac-
tions on Visualization and Computer Graphics 23.1 (Jan. 2017), 541–
550. DOI:10.1109/TVCG.2016.2598958 18.
[BS15] BOUTS, QUIR IJN W. and SP ECK MAN N, BE TTI NA. “Clustered
edge routing”. 2015 IEEE Pacific Visualization Symposium (Paci-
ficVis). Apr. 2015, 55–62. DOI:10 . 1109 / PACIFICVIS . 2015 .
7156356 12,18.
[BST00] BRANDES, UL RIK, SHU BINA , GAL INA, and TAMASSIA,
ROBERTO. “Improving Angular Resolution in Visualizations of Ge-
ographic Networks”. Data Visualization 2000. Ed. by de LEEU W,
WILLEM CORNELIS and van LIERE, ROBERT. Eurographics. Springer
Vienna, 2000, 23–32. ISB N: 978-3-7091-6783-0 10,18.
[BSV11a] BUCHIN, K., S PEC KMANN, B., and VER BEE K, K. “Flow Map
Layout via Spiral Trees”. IEEE Transactions on Visualization and Com-
puter Graphics 17.12 (Dec. 2011), 2536–2544. DOI:10.1109/TVCG.
2011.202 2,6,7,12,13.
[BSV11b] BUCHIN, KE VIN, S PECKMANN, BETT INA, and VERBEEK,
KEVIN. “Angle-Restricted Steiner Arborescences for Flow Map Lay-
out”. Algorithms and Computation. Ed. by AS ANO, TAKAO, NAKANO,
SHIN-ICHI, OK AMOTO, YOSHIO, and WATANABE, OSA MU. Lecture
Notes in Computer Science. Berlin, Heidelberg: Springer, 2011, 250–
259. DOI:10.1007/978-3-642-25591-5_27 6.
[BW98] BRANDES, UL RIK and WAGN ER, DOROTHEA. “Using Graph
Layout to Visualize Train Interconnection Data”. Graph Drawing. Ed.
by WHITESIDES, SU E H. Lecture Notes in Computer Science. Springer
Berlin Heidelberg, 1998, 44–56. DOI:10.1142/9789812794741_
0009 7,10,18.
[CEH96] COX, K ENNETH C., EICK, STEPHEN G., and HE, TAOS ONG .
“3D Geographic Network Displays”. SIGMOD Rec. 25.4 (Dec.
1996), 50–54. DOI:10.1145/245882.245901 7,13,15,20.
[CKS*16] CORNEL, D., KONE V, A., SADRANSKY, B., et al. “Composite
Flow Maps”. Computer Graphics Forum 35.3 (June 1, 2016), 461–470.
DOI:10.1111/cgf.12922 11,15.
[CZQ*08] CUI, W., ZHO U, H., QU, H., et al. “Geometry-Based Edge
Clustering for Graph Visualization”. IEEE Transactions on Visualiza-
tion and Computer Graphics 14.6 (Nov. 2008), 1277–1284. DO I:10 .
1109/TVCG.2008.135 12.
[DLY*05] DOA NTAM PH AN, LING XI AO, YE H, R., et al. “Flow map lay-
out”. IEEE Symposium on Information Visualization, 2005. INFOVIS
2005. Minneapolis, MN, USA: IEEE, 2005, 219–224. DO I:10.1109/
INFVIS.2005.1532150 12,13.
[DSD14a] DEBIASI, AL BERT O, SIMÕES, BRU NO, and DEAMICIS, RAF -
FAEL E. “3DArcLens: A technique for the exploration of geographical
networks”. 2014 IEEE Conference on Visual Analytics Science and Tech-
nology (VAST). Oct. 2014, 245–246. DOI:10. 1109/ VAST .2014 .
7042512 16,18.
[DSD14b] DEBIASI, AL BERT O, SIMÕES, BRU NO, and DEAMICIS, RAF-
FAEL E. “Force Directed Flow Map Layout”. International Conference on
Information Visualization Theory and Applications (Jan. 8, 2014). DOI:
10.5220/0004744601700177 7,12.
[DTH09] DENT, BORDEN D, TORGUSON, JEFF, and HOD LER , THO MAS
W. Cartography: Thematic Map Design. McGraw-Hill Education, 2009.
ISBN: 978-0697384959 3.
[DWCM18] DONG, WEIHUA, WAN G, SHENGKAI, CH EN, YIZHOU, and
MENG, LIQIU. “Using Eye Tracking to Evaluate the Usability of
Flow Maps”. ISPRS International Journal of Geo-Information 7.7 (July
2018), 281. DOI:10.3390/ijgi7070281 16,18.
[Fie18] FIELD, K ENNETH.Cartography: a compendium of design thinking
for mapmakers. Redlands, California: Esri Press, 2018. 549 pp. 3.
[GFC04] GHONIEM, M., FEKETE, J.-D ., and CAS TAGLI OLA, P. “A Com-
parison of the Readability of Graphs Using Node-Link and Matrix-
Based Representations”. IEEE Symposium on Information Visualization.
Austin, TX, USA: IEEE, 2004, 17–24. DOI:10 . 1109 / INFVIS .
2004.1 17.
[GHNS11] GANSNER, EM DEN R., HU, YIFAN, NORTH , STE PHE N, and
SCHEIDEGGER, CARLOS. “Multilevel agglomerative edge bundling for
visualizing large graphs”. 2011 IEEE Pacific Visualization Sympo-
sium. Mar. 2011, 187–194. DOI:10. 1109 / PACIFICVIS . 2011 .
5742389 7,12.
[GRE11] GHANI, S., RICHE, N. HE NRY, and ELMQVIST, N. “Dynamic
Insets for Context-Aware Graph Navigation”. Computer Graphics Forum
30.3 (2011), 861–870. DOI:10 . 1111 / j . 1467 - 8659 . 2011 .
01935.x 8,15,16,18.
[GSRB19] GRASER, AN ITA, SCHMIDT, JOHANNA, ROT H, FLORIAN,
and BRÄNDLE, NO RBE RT. “Untangling origin-destination flows in ge-
ographic information systems”. Information Visualization 18.1 (Jan. 1,
2019), 153–172. DO I:10.1177/1473871617738122 12.
[Guo07] GUO, DIANSHENG. “Visual analytics of spatial interaction pat-
terns for pandemic decision support”. International Journal of Geo-
graphical Information Science 21.8 (Sept. 1, 2007), 859–877. DO I:10.
1080/13658810701349037 2,11.
[Guo09] GUO, D. “Flow Mapping and Multivariate Visualization of Large
Spatial Interaction Data”. IEEE Transactions on Visualization and Com-
puter Graphics 15.6 (Nov. 2009), 1041–1048. DOI:10.1109/TVCG.
2009.143 9,13,15.
[GZ14] GUO, D. and ZH U, X. “Origin-Destination Flow Data Smooth-
ing and Mapping”. IEEE Transactions on Visualization and Computer
Graphics 20.12 (Dec. 2014), 2043–2052. DOI:10 . 1109 / TVCG .
2014.2346271 10,12,18.
[HCC*19] HE, JING, CHEN, HAONA N, CH EN, YIJIN, et al. “Diverse Visu-
alization Techniques and Methods of Moving-Object-Trajectory Data: A
Review”. ISPRS International Journal of Geo-Information 8.2 (Jan. 29,
2019), 63. DO I:10.3390/ijgi8020063 3.
[Hen13] HENNEMANN, STE FAN. “Information-rich visualisation of dense
geographical networks”. Journal of Maps 9.1 (Mar. 1, 2013), 68–75.
DOI:10.1080/17445647.2012.753850 2,8,9,14,17,19.
[HFM07] HENRY, NATHALIE, FEK ETE , JEA N-DANIEL, and MCGUFFIN,
MICHAEL J. “NodeTrix: a Hybrid Visualization of Social Networks”.
IEEE Transactions on Visualization and Computer Graphics 13.6 (Nov.
2007), 1302–1309. DO I:10.1109/TVCG.2007.70582 20.
[HHS20] HOGRÄFER, MARIUS, HEITZLER, MAGNUS, and SCHULZ,
HANS-JÖRG. “The State of the Art in Map-Like Visualization”. Com-
puter Graphics Forum 39.3 (2020), 647–674. DO I:10 . 1111 / cgf .
14031 3,5,7.
[HIvWF11] HOLTEN, DA NNY, ISENBERG, PET RA, van WIJK, JARKE J.,
and FEKETE, JEA N-DANIEL. “An extended evaluation of the readabil-
ity of tapered, animated, and textured directed-edge representations in
node-link graphs”. 2011 IEEE Pacific Visualization Symposium. Mar.
2011, 195–202. DO I:10.1109/PACIFICVIS.2011.5742390 16.
[HMdN06] HONG, SE OK-HEE, MERRICK, DAMIAN, and do NAS CI-
MENTO, HUGO A. D. “Automatic visualisation of metro maps”. Journal
of Visual Languages & Computing 17.3 (June 1, 2006), 203–224. DOI:
10.1016/j.jvlc.2005.09.001 8,18.
[HMM00] HERMAN, I., MELANCON, G., and MARSHALL, M. S. “Graph
visualization and navigation in information visualization: A survey”.
IEEE Transactions on Visualization and Computer Graphics 6.1 (Jan.
2000), 24–43. DO I:10.1109/2945.841119 2.
[HS11] HAUNERT, JA N-HENRIK and SERING, LE ON. “Drawing Road
Networks with Focus Regions”. IEEE Transactions on Visualization and
Computer Graphics 17.12 (Dec. 2011), 2555–2562. DOI:10 . 1109 /
TVCG.2011.191 8.
[HSS11] HADLAK, S., SCHULZ, H., and SCHUMANN, H. “In Situ Ex-
ploration of Large Dynamic Networks”. IEEE Transactions on Visual-
ization and Computer Graphics 17.12 (Dec. 2011), 2334–2343. DOI:
10.1109/TVCG.2011.213 13,15,18.
[HSS15] HADLAK, ST EFFE N, SCHUMANN, HE IDRUN, and SCHULZ,
HANS-JÖRG. “A survey of multi-faceted graph visualization”. Euro-
graphics Conference on Visualization (EuroVis). Vol. 33. The Euro-
graphics Association Cagliary, Italy, 2015, 1–20. DOI:10 . 2312 /
eurovisstar.20151109 3,6,12.
submitted to COMPUTER GRAPHICS Forum (3/2021).

24 S. Schöttler, Y. Yang, H. Pfister & B. Bach / Geospatial Network Visualization
[HTSS10] HADLAK, S., TOMINSKI, C., SCHULZ, H.-J ., and SCHU-
MANN, H. “Visualization of attributed hierarchical structures in a spa-
tiotemporal context”. International Journal of Geographical Informa-
tion Science 24.10 (Oct. 11, 2010), 1497–1513. DO I:10 . 1080 /
13658816.2010.510840 13.
[HvW09a] HOLTEN, DA NNY and van WIJK, JARKE J . “A User Study
on Visualizing Directed Edges in Graphs”. Proceedings of the SIGCHI
Conference on Human Factors in Computing Systems. CHI ’09. New
York, NY, USA: ACM, 2009, 2299–2308. DOI:10.1145/1518701.
1519054 16.
[HvW09b] HOLTEN, DA NNY and van WIJK, JARKE J . “Force-Directed
Edge Bundling for Graph Visualization”. Computer Graphics Forum
28.3 (2009), 983–990. DO I:10 . 1111 / j . 1467 - 8659 . 2009 .
01450.x 11,12,18.
[IHK*17] ISENBERG, P., HEIMERL, F., KOC H, S., et al. “Vispubdata.org:
A Metadata Collection About IEEE Visualization (VIS) Publications”.
IEEE Transactions on Visualization and Computer Graphics 23.9 (Sept.
2017), 2199–2206. DOI:10.1109/TVCG.2016.2615308 4.
[ITH16] IBA RRA, J. C., TRIANA, J. A., and HE RNÁ NDE Z, J. T. “Visu-
alization of origin-destination matrices using a connection barchart and
coordinated maps”. 2016 IEEE Pacific Visualization Symposium (Paci-
ficVis). Apr. 2016, 189–193. DOI:10. 1109/PACIFICVIS .2016 .
7465268 7,13.
[JE12] JAVED, WAQAS and ELMQVI ST, NIKLAS. “Exploring the design
space of composite visualization”. 2012 IEEE Pacific Visualization Sym-
posium. Feb. 2012, 1–8. DOI:10 . 1109 / PacificVis . 2012 .
6183556 12,13.
[JŠA*17] JENNY, BERNHARD, ŠAVRI ˇ
C, BOJAN, ARNOLD, NICHOLAS
D., et al. “A Guide to Selecting Map Projections for World and Hemi-
sphere Maps”. Choosing a Map Projection. Ed. by LAPAINE, M ILJE NKO
and USERY, E. LYNN. Lecture Notes in Geoinformation and Cartog-
raphy. Cham: Springer International Publishing, 2017, 213–228. DOI:
10.1007/978-3- 319-51835-0_9 20.
[JSM*17] JENNY, BERNHARD, STEPHEN, DANIEL M., MUEH LEN HAUS ,
IAN, et al. “Force-directed layout of origin-destination flow maps”. In-
ternational Journal of Geographical Information Science 31.8 (Aug. 3,
2017), 1521–1540. DOI:10.1080 /13658816. 2017.1307378 2,
12,18.
[JSM*18] JENNY, BERNHARD, STEPHEN, DANIEL M., MUEH LEN HAUS ,
IAN, et al. “Design principles for origin-destination flow maps”. Cartog-
raphy and Geographic Information Science 45.1 (Jan. 2, 2018), 62–75.
DOI:10.1080/15230406.2016.1262280 3,12,13,16,18.
[KB16] KAYA, BER KAY and BAL CIS OY, SE LIM. “Multi-Resolution Vi-
sualisation of Geographic Network Traffic”. Augmented Reality, Virtual
Reality, and Computer Graphics. Ed. by DEPAOLI S, LUCIO TOM-
MASO and MONGELLI, AN TON IO. Lecture Notes in Computer Science.
Springer International Publishing, 2016, 52–71. DO I:10.1007/978-
3-319- 40621-3_4 7,10,20.
[KG17] KOYL U, CAG LAR and GUO, DIANSHENG. “Design and evalu-
ation of line symbolizations for origin–destination flow maps”. Infor-
mation Visualization 16.4 (Oct. 1, 2017), 309–331. DOI:10. 1177 /
1473871616681375 16.
[KJW*18] KIM, SEOKYEON, JEONG, SEONGMIN, WOO, INSO O, et al.
“Data Flow Analysis and Visualization for Spatiotemporal Statistical
Data without Trajectory Information”. IEEE Transactions on Visualiza-
tion and Computer Graphics 24.3 (Mar. 2018), 1287–1300. DOI:10 .
1109/TVCG.2017.2666146 10,12,18.
[KK17] KERRAC HER, N. and KENNEDY, J. “Constructing and Evaluating
Visualisation Task Classifications: Process and Considerations”. Com-
puter Graphics Forum 36.3 (June 1, 2017), 47–59. DOI:10 . 1111 /
cgf.13167 21.
[KLSC12] KIM, KAMYOUNG, LEE, SANG-IL, SHIN, JUNGYEOP, and
CHOI, EUN YOUNG. “Developing a Flow Mapping Module in a GIS En-
vironment”. The Cartographic Journal 49.2 (May 1, 2012), 164–175.
DOI:10.1179/174327711X13166800242356 11.
[KSDW13] KEL LY, M., SLINGSBY, A., DYKES, J., and WOOD, J. “His-
torical Internal Migration in Ireland”. GIS Research UK (GISRUK). Liv-
erpool, UK, 2013. UR L:https: // openaccess. city. ac. uk/
id/eprint/2052/ (visited on 10/09/2020) 7.
[KW05] KAPLER, THO MAS and WRIGHT, WILLIAM. “GeoTime In-
formation Visualization”. Information Visualization 4.2 (June 1,
2005), 136–146. DO I:10 . 1057 / palgrave . ivs . 9500097 11,
15,18.
[LBA10a] LAM BERT, A ., BO URQ UI, R., and AUB ER, D. “Winding
Roads: Routing edges into bundles”. Computer Graphics Forum 29.3
(2010), 853–862. DO I:10. 1111/ j. 1467- 8659 .2009 .01700 .
x11,12.
[LBA10b] LAM BERT, A NTO INE, BOU RQUI, ROMAIN, and AUBER,
DAVID. “3D Edge Bundling for Geographical Data Visualization”. 2010
14th International Conference Information Visualisation. 2010 14th In-
ternational Conference Information Visualisation. July 2010, 329–335.
DOI:10.1109/IV.2010.53 7,15.
[LBW17a] LANDESBERGER, T. VON , BRE MM, S., and WUNDERLICH,
M. “Typology of Uncertainty in Static Geolocated Graphs for Visualiza-
tion”. IEEE Computer Graphics and Applications 37.5 (2017), 18–27.
DOI:10.1109/MCG.2017.3621220 11,19.
[LBW17b] LI, CHENHUI, BACIU, GEORGE, and WANG, YUNZHE.
“Module-based visualization of large-scale graph network data”. Jour-
nal of Visualization 20.2 (May 1, 2017), 205–215. DOI:10 . 1007 /
s12650-016- 0375-5 9–11,15,18.
[LHT17a] LHUILLIER, A., HURTER, C., and TELE A, A. “FFTEB: Edge
bundling of huge graphs by the Fast Fourier Transform”. 2017 IEEE Pa-
cific Visualization Symposium (PacificVis). Apr. 2017, 190–199. DOI:
10.1109/PACIFICVIS.2017.8031594 2,12,18.
[LHT17b] LHUILLIER, A., HURTER, C., and TELE A, A. “State of the
Art in Edge and Trail Bundling Techniques”. Computer Graphics Fo-
rum 36.3 (June 2017), 619–645. DOI:10. 1111/ cgf. 13213 4,12,
15.
[LLC*19] LYU, YAN, LIU, XU, CHEN, HAN YI, et al. “OD Morphing: Bal-
ancing Simplicity with Faithfulness for OD Bundling”. IEEE Transac-
tions on Visualization and Computer Graphics (2019), 1–1. DOI:10 .
1109/TVCG.2019.2934657 15.
[LMYH11] LUO, WEI, MACEAC HRE N, AL AN M., YIN, PEIF ENG, and
HARDISTY, FRA NK. “Spatial-social Network Visualization for Ex-
ploratory Data Analysis”. Proceedings of the 3rd ACM SIGSPATIAL In-
ternational Workshop on Location-Based Social Networks. LBSN ’11.
New York, NY, USA: ACM, 2011, 65–68. D OI:10.1145/2063212.
2063216 15.
[LPP*06] LEE, BONGSHIN, PLAISANT, CATHERINE, PARR, CYNTHIA
SIMS, et al. “Task Taxonomy for Graph Visualization”. Proceedings of
the 2006 AVI Workshop on BEyond Time and Errors: Novel Evalua-
tion Methods for Information Visualization. BELIV ’06. Venice, Italy:
Association for Computing Machinery, 2006, 1–5. DOI:10 . 1145 /
1168149.1168168 20,21.
[LWLY15] LU, M IN, WANG, ZUC HAO, LIAN G, JIE, and YUA N, XI-
AORU. “OD-Wheel: Visual design to explore OD patterns of a central
region”. 2015 IEEE Pacific Visualization Symposium (PacificVis). Apr.
2015, 87–91. DO I:10.1109/PACIFICVIS.2015.7156361 13.
[MCH*09] MOSCOV ICH, TOME R, CHEVALIER, FAN NY, HE NRY,
NATHALIE, et al. “Topology-aware Navigation in Large Networks”.
Proceedings of the SIGCHI Conference on Human Factors in Comput-
ing Systems. CHI ’09. New York, NY, USA: ACM, 2009, 2319–2328.
DOI:10.1145/1518701.1519056 2,16,20.
[MG06] MERRICK, DAMIAN and GUDMUNDSSON, JOACHIM. “Increas-
ing the Readability of Graph Drawings with Centrality-based Scaling”.
Proceedings of the 2006 Asia-Pacific Symposium on Information Visu-
alisation - Volume 60. APVis ’06. Darlinghurst, Australia, Australia:
Australian Computer Society, Inc., 2006, 67–76. DOI:10 . 1145 /
1151903.1151914 8,18.
[MGM*19] MCGEE, F., GHONIEM, M., MELANÇON, G., et al. “The
State of the Art in Multilayer Network Visualization”. Computer Graph-
ics Forum 38.6 (2019), 125–149. DO I:10.1111/cgf.13610 2.
submitted to COMPUTER GRAPHICS Forum (3/2021).

S. Schöttler, Y. Yang, H. Pfister & B. Bach / Geospatial Network Visualization 25
[MH17] MCNEILL, GRAHAM and HAL E, SCOTT A. “Generating Tile
Maps”. Computer Graphics Forum 36.3 (June 2017), 435–445. DO I:
10.1111/cgf.13200 8.
[MHCF96] MUNZNER, T., HO FFMA N, E., C LAFF Y, K., and FENNER,
B. “Visualizing the global topology of the MBone”. Proceedings IEEE
Symposium on Information Visualization ’96. Proceedings IEEE Sym-
posium on Information Visualization ’96. Oct. 1996, 85–92. DOI:10 .
1109/INFVIS.1996.559226 7.
[NB13] NOCAJ, ARLIND and BRA NDE S, UL RIK. “Stub Bundling and
Confluent Spirals for Geographic Networks”. Graph Drawing. Ed. by
WISMATH, ST EPH EN and WO LFF, ALEX AND ER. Lecture Notes in
Computer Science. Springer International Publishing, 2013, 388–399.
DOI:10.1007/978-3-319-03841-4_34 12.
[NBD*15] NOCKE, T., BUSCHMANN, S., DONGES, J. F., et al. “Re-
view: visual analytics of climate networks”. Nonlinear Processes in Geo-
physics 22.5 (Sept. 23, 2015), 545–570. DOI:10 . 5194 / npg - 22 -
545-2015 3.
[NMSL19] NOBRE, C ., MEYE R, M., S TREIT, M. , and LEX, A . “The State
of the Art in Visualizing Multivariate Networks”. Computer Graphics
Forum 38.3 (2019), 807–832. DO I:10.1111/cgf.13728 2.
[Nöl09] NÖLLENBURG, MARTIN. “Network Visualization: Algorithms,
Applications, and Complexity”. PhD Thesis. Universität Karlsruhe,
2009. DOI:10.5445/IR/1000011456 7.
[NW11] NÖLLENBURG, M. and WOLFF, A. “Drawing and Labeling
High-Quality Metro Maps by Mixed-Integer Programming”. IEEE
Transactions on Visualization and Computer Graphics 17.5 (May
2011), 626–641. DO I:10.1109/TVCG.2010.81 7.
[OHN*18] OTT EN, HE IKE , HILDEBRAND, LEN NART, NAGEL, TIL L, et
al. “Shifted Maps: Revealing spatio-temporal topologies in movement
data”. 2018 IEEE VIS Arts Program (VISAP). 2018 IEEE VIS Arts Pro-
gram (VISAP). Oct. 2018, 1–10. DO I:10 . 1109 / VISAP45312 .
2018 . 9046054.URL:https : / / shifted - maps . com/ 2,8,
14,15,18,21.
[OJK18] OKO E, MERSHACK, JIANU, RADU, and KOBOUROV, ST EPH EN.
“Revisited Experimental Comparison of Node-Link and Matrix Rep-
resentations”. Graph Drawing and Network Visualization. Vol. 10692.
Cham: Springer International Publishing, 2018, 287–302. DO I:10 .
1007/978-3- 319-73915-1_23 17.
[OJK19] OKO E, MERSHACK, JIANU, RADU, and KOBOUROV, STEPHE N.
“Node-Link or Adjacency Matrices: Old Question, New Insights”. IEEE
Transactions on Visualization and Computer Graphics 25.10 (Oct. 1,
2019), 2940–2952. DOI:10.1109/TVCG.2018.2865940 17.
[Pac19] PACIFI CVIS /APVIS. Visualization Symposium, Pacific (formerly
Asia-Pacific {APVIS}). IEEE Xplore Digital Library. 2019. URL:
https://ieeexplore.ieee.org/xpl/conhome/1001657/
all-proceedings (visited on 07/19/2019) 4.
[PHT15] PEYSAKHOVICH, VSE VOLOD, HU RTER, CHRISTOPHE, and
TELEA, AL EXA NDRU. “Attribute-driven edge bundling for general
graphs with applications in trail analysis”. 2015 IEEE Pacific Visual-
ization Symposium (PacificVis). Apr. 2015, 39–46. DOI:10 . 1109 /
PACIFICVIS.2015.7156354 12.
[PLCP12] PENG, DI CHAO , LU, NE NG, CHEN, WEI, and PENG, QUN-
SHENG. “SideKnot: Revealing relation patterns for graph visualization”.
2012 IEEE Pacific Visualization Symposium. Feb. 2012, 65–72. DO I:
10.1109/PacificVis.2012.6183575 12,15.
[RAB*18] ROM AT, HUGO, APP ERT, CAROLINE, BACH, BEN JAMI N, et
al. “Animated Edge Textures in Node-Link Diagrams: A Design Space
and Initial Evaluation”. Proceedings of the 2018 CHI Conference on Hu-
man Factors in Computing Systems. CHI ’18. New York, NY, USA:
ACM, 2018, 187:1–187:13. DOI:10.1145/3173574.3173761 11,
15,16.
[RDLC12] RICHE, NATHALIE HENRY, DWYER, TIM, LEE, BONGSHIN,
and CARPENDAL E, SHEELAGH. “Exploring the Design Space of Inter-
active Link Curvature in Network Diagrams”. Proceedings of the In-
ternational Working Conference on Advanced Visual Interfaces. AVI
’12. New York, NY, USA: ACM, 2012, 506–513. D OI:10 . 1145 /
2254556.2254652 15,16,18.
[Ren18] RENDGEN, SANDRA.The Minard System: The Complete Statisti-
cal Graphics of Charles-Joseph Minard. Chronicle Books, 2018. ISBN :
978-1-61689-633-1 1,12.
[RM13] ROZ ENB LAT, CÉLINE and MELANÇON, GUY.Methods for mul-
tilevel analysis and visualisation of geographical networks. Springer,
2013. DOI:10.1007/978-94-007-6677-8 3.
[Rob07] ROB ERTS, J ONATHAN C. “State of the Art: Coordinated & Multi-
ple Views in Exploratory Visualization”. Fifth International Conference
on Coordinated and Multiple Views in Exploratory Visualization (CMV
2007). Zurich, Switzerland: IEEE, July 2007, 61–71. DOI:10. 1109/
CMV.2007.20 13.
[Rob95] ROBINSON, ARTH UR HOWAR D, ed. Elements of cartography.
6th ed. New York: Wiley, 1995. 674 pp. ISB N: 978-0-471-55579-7 3.
[Rod05] ROD GER S, PET ER. “Graph Drawing Techniques for Geographic
Visualization”. Exploring Geovisualization. Ed. by DYKES, JASO N,
MACEAC HRE N, AL AN M., and KRA AK, MENN O-JA N. 2005 3.
[Rot13] ROTH, ROBE RT E. “An Empirically-Derived Taxonomy of In-
teraction Primitives for Interactive Cartography and Geovisualization”.
IEEE Transactions on Visualization and Computer Graphics 19.12 (Dec.
2013), 2356–2365. DO I:10.1109/TVCG.2013.130 21.
[SHH11] SELASSIE, D., HELLER, B., and HE ER, J. “Divided Edge
Bundling for Directional Network Data”. IEEE Transactions on Visu-
alization and Computer Graphics 17.12 (Dec. 2011), 2354–2363. DOI:
10.1109/TVCG.2011.190 12.
[SHS11] SCHULZ, H., HADLAK, S., and SC HUM ANN, H. “Point-Based
Visualization for Large Hierarchies”. IEEE Transactions on Visualiza-
tion and Computer Graphics 17.5 (May 2011), 598–611. DOI:10 .
1109/TVCG.2010.89 13.
[SIG] SIGSPATIAL, ACM. The ACM SIGSPATIAL International Con-
ferences on Advances in Geographic Information Systems. ACM
SIGSPATIAL International Conferences. URL:https:/ /dl . acm.
org/conference/gis/proceedings (visited on 06/01/2020) 4.
[SJ17] STEPHEN, DANIEL M. and JENNY, BERNHARD. “Automated lay-
out of origin–destination flow maps: U.S. county-to-county migration
2009–2013”. Journal of Maps 13.1 (Jan. 1, 2017). Licensed under CC
BY 4.0, 46–55. DO I:10.1080/17445647.2017.1313788 9,13.
[SMKH09] SLOCUM, TE RRY A., MCMASTER, ROBERT B., KE SSL ER,
FRITZ C., and HOWAR D, HUGH H. Thematic Cartography and Geovi-
sualization. Pearson Prentice Hall, 2009 3.
[SNG*17] SCHULZ, CH RIS TOP H, NOC AJ, ARLIND, GOERTLER,
JOCHEN, et al. “Probabilistic Graph Layout for Uncertain Network
Visualization”. IEEE Transactions on Visualization and Computer
Graphics 23.1 (Jan. 2017), 531–540. DOI:10. 1109/ TVCG .2016 .
2598919 11,19.
[Sny87] SNYDER, JOHN P. Map projections: A working manual. USGS
Numbered Series 1395. Washington, D.C.: U.S. Government Print-
ing Office, 1987. URL:http : / / pubs . er . usgs . gov /
publication/pp1395 (visited on 05/27/2020) 6,8,20.
[Spr19] SPRINGERLINK.International Symposium on Graph Draw-
ing and Network Visualization. SpringerLink. 2019. URL:https :
/ / link . springer . com / conference / gd (visited on
07/19/2019) 4.
[Sun19] SUN, SHI PEN G. “A spatial one-to-many flow layout algorithm us-
ing triangulation, approximate Steiner trees, and path smoothing”. Car-
tography and Geographic Information Science 46.3 (May 4, 2019), 243–
259. DOI:10.1080/15230406.2018.1437359 11–13.
[SV10] SPECKMANN, BET TINA and VER BEE K, KEVIN. “Necklace
Maps”. IEEE Transactions on Visualization and Computer Graphics
16.6 (Nov. 2010), 881–889. DOI:10. 1109 / TVCG . 2010 . 180 9,
17,20.
[SV15] SPECKMANN, BET TINA and VER BEE K, KEV IN. “Algorithms for
Necklace Maps”. International Journal of Computational Geometry &
Applications 25.01 (Mar. 2015). Publisher: World Scientific Publishing
Co., 15–36. DOI:10.1142/s021819591550003x 9.
submitted to COMPUTER GRAPHICS Forum (3/2021).

26 S. Schöttler, Y. Yang, H. Pfister & B. Bach / Geospatial Network Visualization
[TAvHS06] TOMINSKI, C., ABE LLO , J., van HAM , F., and SCHU MANN,
H. “Fisheye Tree Views and Lenses for Graph Visualization”. Tenth In-
ternational Conference on Information Visualisation (IV’06). Tenth In-
ternational Conference on Information Visualisation (IV’06). London,
England: IEEE, 2006, 17–24. DO I:10.1109/IV.2006.54 18.
[Tob87] TOBLER, WALD O R. “Experiments in migration mapping by
computer”. The American Cartographer 14.2 (1987), 155–163. DO I:
10.1559/152304087783875273 10.
[VBW15] VEHLOW, CORINNA, BECK, FABIAN, and WEI SKOPF,
DANIEL. “The State of the Art in Visualizing Group Structures in
Graphs”. Eurographics Conference on Visualization (EuroVis) - STARs.
Ed. by BO RGO, R., GA NOVELLI, F., and VI OLA, I. The Eurographics
Association, 2015. DOI:10.2312/eurovisstar.20151110 2.
[vdEvW14] Van den ELZ EN, STEF and van WIJK, JARKE J . “Multivari-
ate Network Exploration and Presentation: From Detail to Overview via
Selections and Aggregations”. IEEE Transactions on Visualization and
Computer Graphics 20.12 (Dec. 2014), 2310–2319. DOI:10 . 1109 /
TVCG.2014.2346441 9–11,15,17.
[vDL19] Van DIJK, THOMAS C. and L ÖFFLE R, AN DRE. “Practical Topo-
logically Safe Rounding of Geographic Networks”. Proceedings of the
27th ACM SIGSPATIAL International Conference on Advances in Ge-
ographic Information Systems. SIGSPATIAL ’19: 27th ACM SIGSPA-
TIAL International Conference on Advances in Geographic Information
Systems. Chicago IL USA: ACM, Nov. 5, 2019, 239–248. DOI:10 .
1145/3347146.3359347 8.
[VFAA17] VROTSOU, KATERINA, FUCHS, GEOR G, AN DRI ENKO, NA-
TALI A, and AN DRI ENKO, GEN NADY. “An Interactive Approach for Ex-
ploration of Flows Through Direction-Based Filtering”. Journal of Geo-
visualization and Spatial Analysis 1.1 (May 9, 2017). Licensed under CC
BY 4.0.DOI:10.1007/s41651-017-0001-7 10,11,15.
[vLKS*11] Von LANDESBERGER, TATIANA, KUIJPER, AR JAN,
SCHRECK, TOBIAS, et al. “Visual Analysis of Large Graphs:
State-of-the-Art and Future Research Challenges”. Comput. Graph.
Forum 30 (Sept. 1, 2011), 1719–1749. DO I:10 . 1111 / j . 1467 -
8659.2011.01898.x 2.
[WC07] WONG, NE LSO N and CAR PEN DALE , SHE ELAG H. “Supporting
Interactive Graph Exploration Using Edge Plucking”. Proceedings of
IS&T/SPIE 19th Annual Symposium on Electronic Imaging: Visualiza-
tion and Data Analysis. SPIE and IS&T. 2007. DOI:10.1.1.230.
7985 16.
[WC11] WANG, Y. and CHI , M. “Focus+Context Metro Maps”. IEEE
Transactions on Visualization and Computer Graphics 17.12 (Dec.
2011), 2528–2535. DOI:10.1109/TVCG.2011.205 8,16.
[WCG03] WON G, N., CARPENDALE, S., and GREE NBE RG, S. “Edge-
lens: an interactive method for managing edge congestion in graphs”.
IEEE Symposium on Information Visualization 2003 (IEEE Cat.
No.03TH8714). IEEE Symposium on Information Visualization 2003
(IEEE Cat. No.03TH8714). Oct. 2003, 51–58. DO I:10 . 1109 /
INFVIS.2003.1249008 2,15,16,20.
[WDS10] WOO D, JO, DY KES, JASON, and SLINGSBY, AI DAN. “Visual-
isation of Origins, Destinations and Flows with OD Maps”. The Car-
tographic Journal 47.2 (May 1, 2010), 117–129. DOI:10 . 1179 /
000870410X12658023467367 7–9,13,18.
[Whe15] WHEELER, ANDREW. “Visualization techniques for journey to
crime flow data”. Cartography and Geographic Information Science
42.2 (2015), 239–255. DO I:10.1080/15230406.2014.890545 3.
[Wil19] WILEY ONLINE LIB RARY.Computer Graphics Forum: List of
Issues - Wiley Online Library. Wiley Online Library. 2019. URL:http:
/ / onlinelibrary . wiley . com / loi / 14678659 (visited on
07/19/2019) 4.
[WNT*20] WU, HSIANG-YUN, NIE DER MAN N, BENJAMIN, TAKA-
HASHI, SHIGEO, et al. “A Survey on Transit Map Layout - from De-
sign, Machine, and Human Perspectives”. Computer Graphics Forum
3.39 (2020), 619–646. DOI:10.1111/cgf.14030 3,17.
[Wol07] WOLFF, ALEX AND ER. “Drawing Subway Maps: A Survey”. In-
formatik - Forschung und Entwicklung 22.1 (Dec. 1, 2007), 23–44. DOI :
10.1007/s00450-007- 0036-y 3.
[Wol13] WOLFF, ALEX AND ER. “Graph Drawing and Cartography”.
Handbook of Graph Drawing and Visualization. Ed. by TAMASSIA,
ROBERTO. Chapman and Hall/CRC, 2013 3.
[WTLY12] WU, HSIANG-YU N, TAKAHASHI, SHIGEO, LIN, CHUN -
CHENG, and YEN, HS U-CHUN . “Travel-Route-Centered Metro Map
Layout and Annotation”. Computer Graphics Forum 31.3 (2012), 925–
934. DOI:10.1111/j.1467-8659.2012.03085.x 15.
[WvdWvW09] WILLEMS, NI ELS, van de WETERING, HUU B, and van
WIJK, JARK E J. “Visualization of vessel movements”. Computer Graph-
ics Forum 28.3 (2009), 959–966. DO I:10 .1111 /j . 1467- 8659 .
2009.01440.x 4.
[WWK00] WANG BAL DON ADO, M ICHELLE Q., WOOD RUFF, ALLI SON,
and KUCHINSKY, AL LAN. “Guidelines for using multiple views in in-
formation visualization”. en. Proceedings of the working conference
on Advanced visual interfaces - AVI ’00. Palermo, Italy: ACM Press,
2000, 110–119. IS BN: 978-1-58113-252-6. D OI:10. 1145/345513 .
345271 13.
[XC09] XIAO, NIN GCH UAN and CHUN, YONG WAN. “Visualizing Mi-
gration Flows Using Kriskograms”. Cartography and Geographic In-
formation Science 36.2 (Jan. 1, 2009), 183–191. DO I:10 . 1559 /
152304009788188763 8,9,14,17,19.
[YDGM17] YANG, Y., DWYER, T., GOODWIN, S., and MA RRI OTT,
K. “Many-to-Many Geographically-Embedded Flow Visualisation: An
Evaluation”. IEEE Transactions on Visualization and Computer Graph-
ics 23.1 (Jan. 2017), 411–420. DO I:10 . 1109 / TVCG . 2016 .
2598885 2,11,13,17,18,20.
[YDJ*19] YANG, Y., DWYER, T., JENNY, B., et al. “Origin-Destination
Flow Maps in Immersive Environments”. IEEE Transactions on Visual-
ization and Computer Graphics 25.1 (Jan. 2019), 693–703. DOI:10 .
1109/TVCG.2018.2865192 2,6,7,10,13,15,16,20.
[YG19] YANG, YALONG and GOODWIN, SARAH. “What-Why Analysis
of Expert Interviews: Analysing Geographically-Embedded Flow Data”.
2019 IEEE Pacific Visualization Symposium (PacificVis). Bangkok,
Thailand: IEEE, Apr. 2019, 122–126. DOI:10.1109/PacificVis.
2019.00022 21.
[YJD*18] YANG, YALONG, JEN NY, BERNHARD, DWYER, TIM , et al.
“Maps and Globes in Virtual Reality”. Computer Graphics Forum 37.3
(June 2018), 427–438. IS SN: 01677055. D OI:10 . 1111 / cgf .
13431 20.
[YWZ*19] YAO, XIN, WU, LUN, ZHU, DI, et al. “Visualizing spatial in-
teraction characteristics with direction-based pattern maps”. Journal of
Visualization 22.3 (June 1, 2019), 555–569. DOI:10.1007/s12650-
018-00543- 4 10,12,13,18.
[ZG14] ZHU, XIand GU O, DIANSHENG. “Mapping Large Spatial
Flow Data with Hierarchical Clustering”. Transactions in GIS 18.3
(2014), 421–435. DO I:10.1111/tgis.12100 10,12,18.
[Zou16] ZOU, LINGBO. “A Dynamic Approach for Visualizing Local
and Global Information in Geo-spatial Network Visualizations”. Thesis.
Apr. 29, 2016. URL:https: // DalSpace .library . dal. ca /
/handle/10222/71580 (visited on 07/12/2019) 18.
[ZPYQ13] ZHOU, H., PANPAN XU, YUAN, X., and QU, H. “Edge
bundling in information visualization”. Tsinghua Science and Technol-
ogy 18.2 (Apr. 2013), 145–156. DOI:10 . 1109 / TST . 2013 .
6509098 2,12.
[ZZLL18] ZHANG, MENG -JI A, ZH ANG, K ANG, LI, JI E, and LI, YI-NA.
“Visual Exploration of 3D Geospatial Networks in a Virtual Reality En-
vironment”. The Computer Journal 61.3 (Mar. 1, 2018), 447–458. DOI:
10.1093/comjnl/bxx117 7.
submitted to COMPUTER GRAPHICS Forum (3/2021).